1 2017-07-17 00:15:50	0|sipa|gmaxwell: we likely want that to be dynamic, based on the chosen implementation
  2 2017-07-17 00:16:47	0|gmaxwell|I don't think always buffering two would hurt, other than a slight latency hit.
  3 2017-07-17 00:17:00	0|gmaxwell|8 would be kinda nuts
  4 2017-07-17 00:17:09	0|sipa|well, we can benchmark
  5 2017-07-17 00:19:45	0|gmaxwell|I could be wrong about it mattering at all too. Though I don't think so.
  6 2017-07-17 01:00:13	0|sipa|hmm, i wonder why i think that sse 4.2 is needed for that asm code
  7 2017-07-17 01:00:20	0|sipa|it seems to only use sse 4.1 instructions
  8 2017-07-17 01:00:29	0|sipa|crc32 needs sse 4.2
  9 2017-07-17 02:26:40	0|bitcoin-git|[13bitcoin] 15MarcoFalke pushed 4 new commits to 06master: 02https://github.com/bitcoin/bitcoin/compare/3895e25a7736...bf0a08be281d
 10 2017-07-17 02:26:41	0|bitcoin-git|13bitcoin/06master 14e7a2181 15John Newbery: [wallet] fix zapwallettxes interaction with persistent mempool...
 11 2017-07-17 02:26:41	0|bitcoin-git|13bitcoin/06master 14ff7365e 15John Newbery: [tests] fix flake8 warnings in zapwallettxes.py
 12 2017-07-17 02:26:42	0|bitcoin-git|13bitcoin/06master 144c3b538 15John Newbery: [logs] fix zapwallettxes startup logs
 13 2017-07-17 02:26:56	0|bitcoin-git|[13bitcoin] 15MarcoFalke closed pull request #10330: [wallet] fix zapwallettxes interaction with persistent mempool (06master...06zapwallettxes) 02https://github.com/bitcoin/bitcoin/pull/10330
 14 2017-07-17 04:18:08	0|bitcoin-git|[13bitcoin] 15jnewbery closed pull request #10802: [tests] skip zapwallettxes.py (06master...06skip_zapwallettxes) 02https://github.com/bitcoin/bitcoin/pull/10802
 15 2017-07-17 04:25:40	0|gmaxwell|sipa: you broke the sse4 thing by renaming incompletely. fixy fixy.
 16 2017-07-17 04:37:14	0|sipa|gmaxwell: fixed
 17 2017-07-17 04:42:28	0|cfields|jeez, qt 5.9 made a mess of _everything_
 18 2017-07-17 04:44:06	0|gmaxwell|What did they do
 19 2017-07-17 04:46:55	0|cfields|they're working on modularizing their source/build. but the result (so far) is just a new tangled mess of dependencies. And no working configs without patches
 20 2017-07-17 04:47:39	0|cfields|want cocoa for osx? you'll need a printer of course. And obviously printers require opengl...
 21 2017-07-17 04:47:49	0|luke-jr|O.o
 22 2017-07-17 04:47:51	0|sipa|any reason why we'd really want 5.9?
 23 2017-07-17 04:48:03	0|sipa|cfields: they're 3D printers perhaps? :D
 24 2017-07-17 04:48:45	0|cfields|sipa: lol
 25 2017-07-17 04:49:14	0|cfields|sipa: ironically i was looking forward to 5.9 and its slimmer build
 26 2017-07-17 04:49:16	0|gmaxwell|they probably need opengl for 'lpad()'...
 27 2017-07-17 04:50:01	0|sipa|or for is_integer
 28 2017-07-17 04:50:39	0|cfields|but after messing with it all day, it's not offering many advantages so far
 29 2017-07-17 04:51:01	0|cfields|(other than providing me a stack of patches to upstream)
 30 2017-07-17 04:52:21	0|cfields|so no, i don't think it makes sense to bump
 31 2017-07-17 04:52:41	0|cfields|as-is, we don't build with 5.9. But I haven't seen a bug report yet, so I'm assuming distros aren't shipping it
 32 2017-07-17 04:52:57	0|cfields|(and it'll be a pain to support once they do :( )
 33 2017-07-17 04:53:45	0|luke-jr|Qt used to care about backward compat :/
 34 2017-07-17 04:54:02	0|luke-jr|I noticed some stuff I have won't build with 5.7 because it requires C++11 >_<
 35 2017-07-17 04:54:03	0|cfields|luke-jr: i suspect all is fine if you use qmake
 36 2017-07-17 04:54:13	0|luke-jr|ah
 37 2017-07-17 04:54:23	0|cfields|yea, c++11 became a hard dep for 5.7 i believe
 38 2017-07-17 04:54:26	0|luke-jr|well, FWIW, Gentoo doesn't have 5.8 as even an option yet
 39 2017-07-17 04:54:47	0|cfields|but stl is still optional i think :p
 40 2017-07-17 04:54:50	0|luke-jr|hard dep makes sense; but 5.7 won't allow Qt programs to use non-C++11 :p
 41 2017-07-17 04:55:26	0|sipa|gcc 6 by default already compiles with c++14
 42 2017-07-17 04:55:27	0|luke-jr|anyhow, so I changed my USE flags to build that app for Qt4 <.<
 43 2017-07-17 04:55:37	0|cfields|mm, that's annoying. but kinda makes sense from a support standpoint
 44 2017-07-17 05:58:46	0|wumpus|any reason to not just stick with our current qt for 0.15? that sounds like a nightmare, better let them sort out their problems upstream and not be a guinea pig
 45 2017-07-17 05:59:47	0|wumpus|from a user PoV there seems to be no difference between different qt version for a long time now, as we basically only use the 4.x API
 46 2017-07-17 06:06:17	0|luke-jr|they don't even have a Qt5.9 dance, so..
 47 2017-07-17 06:06:47	0|wumpus|hehe
 48 2017-07-17 06:12:24	0|wumpus|disabling -getinfo on the day of the feature freeze, really?
 49 2017-07-17 06:13:28	0|luke-jr|meh, why not. there's an option to get it back
 50 2017-07-17 06:13:33	0|wumpus|I don't get it, what is the sudden rush?
 51 2017-07-17 06:14:01	0|wumpus|sure, but we could have done that 100 times during 0.15's release cycle
 52 2017-07-17 06:14:08	0|wumpus|but somehow it has to be added in the last day
 53 2017-07-17 06:14:12	0|gmaxwell|leeroy jenkins!
 54 2017-07-17 06:14:19	0|luke-jr|I suppose it adds a new string
 55 2017-07-17 06:14:27	0|luke-jr|(for the -help)
 56 2017-07-17 06:15:19	0|gmaxwell|I think it should be done at the beginning of a cycle. It's bound to cause severe irritation for people who use it at the CLI and we will predictably get PRs to improve other things to compensate.
 57 2017-07-17 06:15:43	0|wumpus|gmaxwell: heh indeed, in this way we don't even get used to it ourself
 58 2017-07-17 06:15:52	0|wumpus|before exposing it to end users
 59 2017-07-17 06:16:14	0|wumpus|can't we just add a comment to the getinfo output instead?
 60 2017-07-17 06:16:23	0|wumpus|Warning: this will be deprecated!
 61 2017-07-17 06:16:27	0|gmaxwell|(or perhaps I'm alone in continuing to use getinfo myself? :) )
 62 2017-07-17 06:16:43	0|wumpus|instead of the add a command line argument dance
 63 2017-07-17 06:16:55	0|gmaxwell|"nag": "This RPC will be deprecated."
 64 2017-07-17 06:17:22	0|wumpus|gmaxwell: I officially stopped using it at least, replaced it with a client-side script that shows more useful information
 65 2017-07-17 06:17:51	0|luke-jr|wumpus: +1, maybe even put it in the warnings software already should look at?
 66 2017-07-17 06:18:07	0|wumpus|getinfo's output is pretty meh, do you ever lok at anything besides the number of connections or the wallet balance?
 67 2017-07-17 06:18:34	0|wumpus|luke-jr: not sure that's a good idea
 68 2017-07-17 06:18:44	0|wumpus|luke-jr: it's not part of the node-wide warnings
 69 2017-07-17 06:19:29	0|luke-jr|not sure why that'd be a reason not to do it
 70 2017-07-17 06:19:45	0|wumpus|because it's a false alarm
 71 2017-07-17 06:19:52	0|wumpus|it's not actually an alarm
 72 2017-07-17 06:20:08	0|luke-jr|"your software is about to break" isn't? ;)
 73 2017-07-17 06:20:24	0|gmaxwell|Connections, block count, balance, warings. And unfortunately, to use seperate rpcs to get this info is three rpcs (four?) and most of them each give a full screen full of information each. :)
 74 2017-07-17 06:20:55	0|luke-jr|getblockcount and getbalance are scalars
 75 2017-07-17 06:21:28	0|sipa|is there any rpc besides getinfo that actually lists the warning?
 76 2017-07-17 06:21:29	0|wumpus|what the hell
 77 2017-07-17 06:21:40	0|wumpus|BlueMatt concerned about atomicity in #8843?!
 78 2017-07-17 06:21:41	0|gmaxwell|true, though I was using getblockchaininfo and getwalletinfo as replacements.
 79 2017-07-17 06:21:42	0|gribble|https://github.com/bitcoin/bitcoin/issues/8843 | rpc: Handle `getinfo` client-side in bitcoin-cli w/ `-getinfo` by laanwj · Pull Request #8843 · bitcoin/bitcoin · GitHub
 80 2017-07-17 06:22:08	0|wumpus|I don't...
 81 2017-07-17 06:22:19	0|luke-jr|gmaxwell: "don't do that"? :p
 82 2017-07-17 06:22:24	0|bitcoin-git|[13bitcoin] 15laanwj closed pull request #8843: rpc: Handle `getinfo` client-side in bitcoin-cli w/ `-getinfo` (06master...062016_09_getinfo_clientside) 02https://github.com/bitcoin/bitcoin/pull/8843
 83 2017-07-17 06:22:36	0|wumpus|never mind, going to put this discussion on the backburner, there's more important things to worry about
 84 2017-07-17 06:22:41	0|wumpus|what needs to get in today?
 85 2017-07-17 06:23:48	0|luke-jr|BIP148, but I assume if the hold-outs on that had changed their mind, they'd say so :/ so probably nothing to do / discuss there
 86 2017-07-17 06:24:13	0|wumpus|no no no no no
 87 2017-07-17 06:24:23	0|sipa|#10831 would be nice, together with #10830 (which i hope can go in s a bugfix later)
 88 2017-07-17 06:24:24	0|gribble|https://github.com/bitcoin/bitcoin/issues/10831 | Batch flushing operations to the walletdb during top up and increase keypool size. by gmaxwell · Pull Request #10831 · bitcoin/bitcoin · GitHub
 89 2017-07-17 06:24:25	0|gribble|https://github.com/bitcoin/bitcoin/issues/10830 | [WIP] [wallet] keypool restore by jnewbery · Pull Request #10830 · bitcoin/bitcoin · GitHub
 90 2017-07-17 06:26:31	0|wumpus|I'll just maintain my own client-side info scripts, I'm not going to try to introduce convenient info functionality in bitcoin-cli anymore
 91 2017-07-17 06:26:39	0|wumpus|this is just too bizarry
 92 2017-07-17 06:26:41	0|achow101|re getinfo: can we add something to the errors field warning about it being removed soon
 93 2017-07-17 06:26:50	0|luke-jr|wumpus: put it in contrib/?
 94 2017-07-17 06:26:58	0|wumpus|luke-jr: no, people can write their own
 95 2017-07-17 06:27:22	0|wumpus|luke-jr: would probably get 100 comments again from people that want something slightly differetn
 96 2017-07-17 06:27:31	0|luke-jr|sigh
 97 2017-07-17 06:27:31	0|wumpus|luke-jr: just tired of it
 98 2017-07-17 06:27:55	0|gmaxwell|please don't add things to the errors field.
 99 2017-07-17 06:28:00	0|wumpus|that's what I said
100 2017-07-17 06:28:06	0|wumpus|just add an extra 'nag' field or so
101 2017-07-17 06:28:12	0|gmaxwell|yes, that would be okay.
102 2017-07-17 06:28:38	0|wumpus|errors would be a horrible place to add it, because it's not a block chain error
103 2017-07-17 06:28:44	0|sipa|agree
104 2017-07-17 06:29:08	0|sipa|there are more urgent things
105 2017-07-17 06:29:24	0|sipa|#10706 should get review
106 2017-07-17 06:29:26	0|gribble|https://github.com/bitcoin/bitcoin/issues/10706 | Improve wallet fee logic and fix GUI bugs by morcos · Pull Request #10706 · bitcoin/bitcoin · GitHub
107 2017-07-17 06:30:01	0|sipa|and #10829 or #10650
108 2017-07-17 06:30:02	0|gribble|https://github.com/bitcoin/bitcoin/issues/10829 | Simple, backwards compatible RPC multiwallet support. by ryanofsky · Pull Request #10829 · bitcoin/bitcoin · GitHub
109 2017-07-17 06:30:06	0|gribble|https://github.com/bitcoin/bitcoin/issues/10650 | Multiwallet: add RPC endpoint support by jonasschnelli · Pull Request #10650 · bitcoin/bitcoin · GitHub
110 2017-07-17 06:31:23	0|gmaxwell|sipa: I was just finishing testing/reviewing 10706.
111 2017-07-17 06:31:56	0|bitcoin-git|[13bitcoin] 15luke-jr closed pull request #10074: Block size/weight fraud proofs (06master...06sizefp) 02https://github.com/bitcoin/bitcoin/pull/10074
112 2017-07-17 07:10:24	0|luke-jr|anyone have anything urgent / must be done before August I can help with for the next hour?
113 2017-07-17 07:26:05	0|bitcoin-git|[13bitcoin] 15laanwj pushed 7 new commits to 06master: 02https://github.com/bitcoin/bitcoin/compare/bf0a08be281d...6859ad2936bf
114 2017-07-17 07:26:06	0|bitcoin-git|13bitcoin/06master 1403ee701 15Alex Morcos: Refactor to use CoinControl in GetMinimumFee and FeeBumper...
115 2017-07-17 07:26:06	0|bitcoin-git|13bitcoin/06master 14ecd81df 15Alex Morcos: Make CoinControl a required argument to CreateTransaction
116 2017-07-17 07:26:07	0|bitcoin-git|13bitcoin/06master 141983ca6 15Alex Morcos: Use CoinControl to pass custom fee setting from QT....
117 2017-07-17 07:26:29	0|bitcoin-git|[13bitcoin] 15laanwj closed pull request #10706: Improve wallet fee logic and fix GUI bugs (06master...06improveWalletFeeLogic) 02https://github.com/bitcoin/bitcoin/pull/10706
118 2017-07-17 07:30:09	0|wumpus|anyone opposed to adding #10829 as a last-ditch try to still have RPC multiwallet support in 0.15?
119 2017-07-17 07:30:10	0|gribble|https://github.com/bitcoin/bitcoin/issues/10829 | Simple, backwards compatible RPC multiwallet support. by ryanofsky · Pull Request #10829 · bitcoin/bitcoin · GitHub
120 2017-07-17 07:30:46	0|wumpus|it is simple to review at least
121 2017-07-17 07:34:13	0|gmaxwell|I support it.
122 2017-07-17 07:49:59	0|bitcoin-git|13bitcoin/06master 141cc251f 15Patrick Strateman: Explicitly search for bdb5.3....
123 2017-07-17 07:49:59	0|bitcoin-git|[13bitcoin] 15laanwj pushed 2 new commits to 06master: 02https://github.com/bitcoin/bitcoin/compare/6859ad2936bf...91edda8f3c81
124 2017-07-17 07:50:00	0|bitcoin-git|13bitcoin/06master 1491edda8 15Wladimir J. van der Laan: Merge #10803: Explicitly search for bdb5.3....
125 2017-07-17 07:50:26	0|bitcoin-git|[13bitcoin] 15laanwj closed pull request #10803: Explicitly search for bdb5.3. (06master...062017-07-02-bdb5.3) 02https://github.com/bitcoin/bitcoin/pull/10803
126 2017-07-17 07:54:20	0|gmaxwell|so on #10831  I'm ambivlent to the performance improvement; I created the PR because I understood that performance issue to be a blocker for increasing the default keypool size, which I think is really important to get out due to bad interactions with hd wallet and rescan.
127 2017-07-17 07:54:22	0|gribble|https://github.com/bitcoin/bitcoin/issues/10831 | Batch flushing operations to the walletdb during top up and increase keypool size. by gmaxwell · Pull Request #10831 · bitcoin/bitcoin · GitHub
128 2017-07-17 07:55:22	0|luke-jr|we should probably just do keypool in the background..?
129 2017-07-17 07:55:50	0|bitcoin-git|[13bitcoin] 15practicalswift closed pull request #10847: Enable devirtualization opportunities by using the final specifier (C++11) (06master...06devirtualization) 02https://github.com/bitcoin/bitcoin/pull/10847
130 2017-07-17 07:56:12	0|wumpus|well, performance improvement or not, doing the keypool top-up in a single db transaction makes sense
131 2017-07-17 07:56:55	0|luke-jr|doing it in a single dbtx would necessitate blocking on it, though, no?
132 2017-07-17 07:57:08	0|luke-jr|at least before any of the keys can be used
133 2017-07-17 07:57:08	0|sipa|?
134 2017-07-17 07:57:09	0|wumpus|especially on VMs with slow sync (most of them), it's super slow as it is right now
135 2017-07-17 07:57:19	0|wumpus|blocking on *what*? key generation is so fast
136 2017-07-17 07:57:58	0|sipa|if it weren't for db syncing, we can generate 10000 per second or so
137 2017-07-17 07:58:05	0|wumpus|the only reason it takes noticable time right now is because of the sync per key
138 2017-07-17 07:58:09	0|gmaxwell|luke-jr: the topup is already a single operation that holds the relevant locks the whole time.
139 2017-07-17 07:58:09	0|luke-jr|i c
140 2017-07-17 07:58:31	0|luke-jr|gmaxwell: yes, I was thinking we shouldn't do that, but otoh, if this is so much of a gain, maybe it doesn't matter
141 2017-07-17 07:58:48	0|wumpus|it's only done once anyhow!
142 2017-07-17 07:58:50	0|gmaxwell|sipa: well not quite that fast, the code paths it goes through per key are very twisty and inefficient, and we aren't multithreaded for it, and we do validate after create. :)
143 2017-07-17 07:59:02	0|wumpus|(at least, so much at once)
144 2017-07-17 07:59:03	0|gmaxwell|but thousands per second sure.
145 2017-07-17 08:00:00	0|wumpus|keypool generation happens a lot during testing though, so #10831 is going to speed up testing
146 2017-07-17 08:00:02	0|gribble|https://github.com/bitcoin/bitcoin/issues/10831 | Batch flushing operations to the walletdb during top up and increase keypool size. by gmaxwell · Pull Request #10831 · bitcoin/bitcoin · GitHub
147 2017-07-17 08:00:45	0|wumpus|(euh.. it would at least if if the number of keys generated stayed the same)
148 2017-07-17 08:00:48	0|gmaxwell|luke-jr: if someone really cared, they could make topup run in batches of no more than X to reduce worst case blocking, it would be a couple lines improvement probably.
149 2017-07-17 08:01:03	0|gmaxwell|wumpus: it'll be faster even with the 1000 size increase, esp on things with slow IO.
150 2017-07-17 08:01:13	0|wumpus|ah yes it's 1000, yes srue
151 2017-07-17 08:01:28	0|wumpus|does it still log a line for every key?
152 2017-07-17 08:01:33	0|wumpus|probably want to get rid of that too
153 2017-07-17 08:01:34	0|luke-jr|well, if it's so fast, it doesn't matter
154 2017-07-17 08:01:40	0|gmaxwell|yes.
155 2017-07-17 08:01:46	0|luke-jr|"added N keys to keypool" ☺
156 2017-07-17 08:03:56	0|gmaxwell|instagibbs reported for 1000 keys (new default) on his presumably SSD equipmt system it went from 20 second to 1.6. so I expect it's still a 20% speedup. We could in the future make many of the tests turn their keypool sizes down to speed them up further.
157 2017-07-17 08:05:17	0|gmaxwell|it would be a trivial (2 loc-ish) change to make it log only one line in the topup.
158 2017-07-17 08:06:09	0|wumpus|should certainly do that, I've always found the per-key message annoying during first run, and it's 10 times as annoying now :)
159 2017-07-17 08:06:14	0|gmaxwell|as in, remove the current line, re-add it two below using the missingInternal missingExternal variables as the count.
160 2017-07-17 08:06:24	0|gmaxwell|okay, should I just add another commit that does that to my PR?
161 2017-07-17 08:06:30	0|sipa|ack
162 2017-07-17 08:06:57	0|wumpus|remove the current line, or move it to debug category if someone theoretically could be interested while troubleshooting
163 2017-07-17 08:07:26	0|gmaxwell|I can't see the reason, it conveys no useful information that the summary wouldn't have.
164 2017-07-17 08:07:40	0|wumpus|agree
165 2017-07-17 08:08:43	0|gmaxwell|and yes, the old one annoyed me too. though leveldb logging with debug on has made me blind to log annoyances... (funny, I did that nice bitfield thing and I don't use it to deactivate leveldb)
166 2017-07-17 08:09:36	0|wumpus|heh yes that's kind of what it was meant for
167 2017-07-17 08:10:31	0|wumpus|though now that it's possible to enable/disable debug bits at runtime, I don't enable any by default anymore
168 2017-07-17 08:11:02	0|wumpus|well, bench. That one is nice.
169 2017-07-17 08:17:30	0|gmaxwell|pushed, not yet tested because it'll take me a half hour to compile it.
170 2017-07-17 08:23:49	0|wumpus|I can test np
171 2017-07-17 08:24:56	0|gmaxwell|I will test, just... when the build finishes.
172 2017-07-17 08:25:28	0|wumpus|half an hour for a single-line change, did you change to rpi for builds? :-)
173 2017-07-17 08:26:47	0|gmaxwell|wumpus: my laptop is slow, and I've switched trees, so its recompiling all the things.
174 2017-07-17 08:41:36	0|wumpus|ccache can help with that a lot
175 2017-07-17 08:43:23	0|gmaxwell|2017-07-17 08:42:25 keypool added 2000 keys (1000 internal), size=2000 (1000 internal)
176 2017-07-17 08:43:25	0|wumpus|though, you need to set the cache size high to survive tree switches
177 2017-07-17 08:43:42	0|gmaxwell|yes, and with a full blockchain on a 256GB disk, I can't do that. :(
178 2017-07-17 08:44:14	0|gmaxwell|okay, only took me 25 minutes.
179 2017-07-17 08:45:19	0|wumpus|2017-07-17 08:45:03 Performing wallet upgrade to 60000
180 2017-07-17 08:45:19	0|wumpus|2017-07-17 08:45:05 keypool added 2000 keys (1000 internal), size=2000 (1000 internal)
181 2017-07-17 08:45:22	0|wumpus|two seconds woohoo
182 2017-07-17 08:45:57	0|gmaxwell|yea, thats still pretty slow considering how fast the underlying crypto is... but fast enough.
183 2017-07-17 08:52:15	0|wumpus|well it's short enough not to be noticable in the overall process, could always be optimzied further if anyone cares
184 2017-07-17 08:52:47	0|gmaxwell|if we later change it to 10k it might be worth it.
185 2017-07-17 08:53:21	0|wumpus|184s for 10k keys here, yea that's definitely too slow
186 2017-07-17 08:53:34	0|wumpus|sorry - 100k keys
187 2017-07-17 08:54:06	0|wumpus|but 18.4 seconds would also be too slow
188 2017-07-17 08:54:46	0|gmaxwell|it could also be broken up and run in the background, I didn't push for 10k for a different reason: bloats up the wallet currently.
189 2017-07-17 08:55:01	0|gmaxwell|esp if you use encryption and immediately mark all the keys you just generated as used.
190 2017-07-17 08:55:01	0|wumpus|then again you had a good point about the encryption
191 2017-07-17 08:55:05	0|wumpus|yes
192 2017-07-17 08:55:43	0|gmaxwell|but we can fix this later by not even creating the wallet until you do something like hit getnewaddress or do encryption.
193 2017-07-17 08:55:52	0|wumpus|there could be some kind of shortcut - if you *never* dealt out a key before encrypting
194 2017-07-17 08:56:06	0|wumpus|then it could just start over with an encrypted wallet
195 2017-07-17 08:56:25	0|gmaxwell|or that.. and also, by using smaller records for keypool entries.
196 2017-07-17 08:56:39	0|wumpus|for multiwallet it'd also be nice to be able to create wallets in encrypted mode
197 2017-07-17 08:57:27	0|wumpus|yes
198 2017-07-17 09:05:56	0|gmaxwell|I think we should generally consider defered init of the wallet because it would allow us to better integrate mandatory backup/recovery.
199 2017-07-17 09:07:08	0|gmaxwell|e.g. in the GUI at least you don't do a wallet unless you complete a series of slightly paternalistic steps that make it likely you actually got some kind of backup.
200 2017-07-17 09:07:54	0|gmaxwell|so: starting with no wallet, and any effort to get an address prompts you to make one, along with a backup of some form.
201 2017-07-17 10:08:18	0|morcos|wumpus thanks for merging 10706.  The last thing I want to nag about is #10707.  It hasn't gotten much review b/c it was dependent on 10706, but its not big.  I think its important in terms of finalizing estimatesmartfee API now that it is no longer marked unstable.  Not sure if you feel the cutoff for it is today.
202 2017-07-17 10:08:20	0|gribble|https://github.com/bitcoin/bitcoin/issues/10707 | Better API for estimatesmartfee RPC by morcos · Pull Request #10707 · bitcoin/bitcoin · GitHub
203 2017-07-17 10:09:18	0|morcos|#10817 would be nice (and isn't yet tagged 0.15) but isn't critical, and so far I haven't heard anyone other than gmaxwell and I's opinion on it.
204 2017-07-17 10:09:19	0|gribble|https://github.com/bitcoin/bitcoin/issues/10817 | Add a discard_rate to avoid small change by morcos · Pull Request #10817 · bitcoin/bitcoin · GitHub
205 2017-07-17 10:14:49	0|bitcoin-git|[13bitcoin] 15jonasschnelli opened pull request #10849: Multiwallet: simplest endpoint support (06master...062017/07/mw_endpoint_simple) 02https://github.com/bitcoin/bitcoin/pull/10849
206 2017-07-17 10:18:15	0|gmaxwell|morcos:  I would not for 10817 that the definition used there is a bit different then the one were thinking of using for the exact matching hurestic, because the dust threshold has a *3 in it.
207 2017-07-17 10:19:13	0|gmaxwell|I think thats okay, but its slightly less of a no-brainer than the factor of one hurestic (where the output is worthless at the long term feerate).
208 2017-07-17 10:24:37	0|morcos|gmaxwell: Yeah I argued a release ago we should remove the *3 from the definition of dust and correspondingly change the dust rate to 3 sat/B .  It would make more sense to talk about everything like that.
209 2017-07-17 10:25:33	0|gmaxwell|morcos: I would agree with that change, makes a lot of sense.
210 2017-07-17 10:26:00	0|gmaxwell|if you make a commit that does that, fix the comment in IsDust which just seems to be inexplicable.
211 2017-07-17 10:26:30	0|gmaxwell|(there is no integer you can multiply 148 by to get 546)
212 2017-07-17 10:26:35	0|morcos|still want to try to do that for 0.15?  or forget it for now
213 2017-07-17 10:26:59	0|morcos|heh!
214 2017-07-17 10:27:13	0|gmaxwell|Well I think you should put it in 10817, other than the 10817 behavior it's a no-op
215 2017-07-17 10:27:13	0|morcos|not forget, but postpone
216 2017-07-17 10:27:26	0|morcos|ok
217 2017-07-17 10:27:53	0|gmaxwell|I'd like to get it in 0.15, but wladimir's call. I think the PR would be better with it, esp as it makes it even a more benign change.
218 2017-07-17 10:28:35	0|gmaxwell|(also, I don't really think we have any other utxo bloat reducers in this release; gotta meet quota)
219 2017-07-17 10:28:49	0|morcos|But then you would make the discard rate still 5 sat/B or you'd make it higher
220 2017-07-17 10:28:58	0|gmaxwell|morcos: the 3x in there was due to the original somewhat stupid formulation that tied it to minimum relay fee.
221 2017-07-17 10:29:06	0|morcos|in the PR I gave the long term fee estimation rate can only serve to LOWER the discard rate
222 2017-07-17 10:29:09	0|morcos|it is not a max
223 2017-07-17 10:29:35	0|morcos|I think it is dangerous to use the existing long term fee estimation rate as a discard rate
224 2017-07-17 10:29:42	0|morcos|it got as high as 100 sat/B a couple months ago
225 2017-07-17 10:30:17	0|gmaxwell|Could be argued that leaving it at 5 is just more conservative and better than not having the feature.
226 2017-07-17 10:31:14	0|gmaxwell|5 means the threshold is about at 1.5 cents at the current prices.
227 2017-07-17 10:31:45	0|morcos|it seems not worth the complication to increase the discard rate from the preexisting 3 sat/B b/c of dust to 5
228 2017-07-17 10:31:50	0|morcos|from 3 to 15 then yes
229 2017-07-17 10:32:38	0|morcos|and i think the fact that it gets MIN'ed with long term fee estimation makes me not worry too much that the 15 will end up biting people in the ass later.  (also it's configurable)
230 2017-07-17 10:33:06	0|gmaxwell|Yes, that also sounds fine to me.
231 2017-07-17 10:33:38	0|morcos|only other question is it's a bit dicey to change the definition of a command line argument like dustrelayfee, but in this case it's a hidden argument that i doubt anyone specified, so its probably ok.
232 2017-07-17 10:33:52	0|morcos|but thats more reason to do it now while it's only been out for one release
233 2017-07-17 10:34:49	0|gmaxwell|it's also good to get this in because the exact matcher code should signficantly increase the number of cases where this happens, so if people are gonna squak about the wallet throwing away a few more cents to avoid change it'll be good to hear about it.
234 2017-07-17 10:35:19	0|morcos|gmaxwell: 546 = 3 * (148 + 34)
235 2017-07-17 10:36:18	0|gmaxwell|ah thats where it comes from; it's including the txout itself.
236 2017-07-17 11:15:54	0|bitcoin-git|13bitcoin/06master 14a8ae0b2 15Dag Robole: Fix resource leak
237 2017-07-17 11:15:54	0|bitcoin-git|[13bitcoin] 15laanwj pushed 2 new commits to 06master: 02https://github.com/bitcoin/bitcoin/compare/91edda8f3c81...8bc6d1f179a0
238 2017-07-17 11:15:55	0|bitcoin-git|13bitcoin/06master 148bc6d1f 15Wladimir J. van der Laan: Merge #10837: Fix resource leak on error in GetDevURandom...
239 2017-07-17 11:16:34	0|bitcoin-git|[13bitcoin] 15laanwj closed pull request #10837: Fix resource leak on error in GetDevURandom (06master...0620170715-fix-leak-1) 02https://github.com/bitcoin/bitcoin/pull/10837
240 2017-07-17 11:20:40	0|morcos|gmaxwell: ok updated 10817 and reduced the discard rate from 15 to 10
241 2017-07-17 11:49:51	0|luke-jr|wumpus: hmm, what about ;wallet=abc (path segment parameter)?
242 2017-07-17 11:49:59	0|wumpus|:-(
243 2017-07-17 11:50:23	0|wumpus|let's pile up even more alternatives
244 2017-07-17 11:50:33	0|wumpus|heh I wonder too
245 2017-07-17 11:50:39	0|luke-jr|wumpus: well, it doesn't have the issue you were concerned with
246 2017-07-17 11:50:59	0|wumpus|maybe we should do something really obscure so that at least no one is happy with it either :p
247 2017-07-17 11:51:23	0|luke-jr|custom HTTP header?
248 2017-07-17 11:51:34	0|wumpus|:-)
249 2017-07-17 11:51:50	0|luke-jr|although that's not so obscure :/
250 2017-07-17 11:51:57	0|wumpus|or a custom field in the JSON-RPC outer structure
251 2017-07-17 11:52:04	0|luke-jr|let's encode it in the Accept header!
252 2017-07-17 11:52:18	0|wumpus|or in the user agent
253 2017-07-17 11:52:25	0|jonasschnelli|:/
254 2017-07-17 11:52:25	0|luke-jr|lol
255 2017-07-17 11:52:43	0|jonasschnelli|lets use sessions and states ... *duck*
256 2017-07-17 11:52:53	0|luke-jr|oooh perfect
257 2017-07-17 11:53:09	0|luke-jr|that would actually be useful for other stuff too
258 2017-07-17 11:53:37	0|luke-jr|historically anyway, maybe not post-named params
259 2017-07-17 11:53:41	0|jonasschnelli|Yes. Would be useful to lose all your coins.. :)
260 2017-07-17 11:53:42	0|jonasschnelli|*loose
261 2017-07-17 11:53:47	0|luke-jr|(I'm thinking settxfee)
262 2017-07-17 11:53:59	0|wumpus|how would that lose coins - store private keys in a session identifier?
263 2017-07-17 11:54:20	0|luke-jr|sounds like bc.i
264 2017-07-17 11:54:37	0|wumpus|"sorry, your session expired" but isntead of losing that long mail, you lose all your coins
265 2017-07-17 11:54:39	0|jonasschnelli|You may loose them because your session with "the other" wallet is still active...
266 2017-07-17 11:54:48	0|jonasschnelli|Though loosing may not directly be possible..
267 2017-07-17 11:54:53	0|jonasschnelli|But sessions is the worst for APIS
268 2017-07-17 11:55:30	0|wumpus|I'm sure someone somewhere thinks it's a good idea
269 2017-07-17 11:55:56	0|jonasschnelli|hehe
270 2017-07-17 11:55:58	0|jonasschnelli|Me too
271 2017-07-17 11:56:06	0|luke-jr|Coinbase
272 2017-07-17 11:56:24	0|jonasschnelli|Does their API work keep state in sessions?
273 2017-07-17 11:56:25	0|luke-jr|OAuth2 is kinda inherently session-based tho.
274 2017-07-17 11:56:40	0|luke-jr|not state, I think
275 2017-07-17 11:56:43	0|jonasschnelli|For Auth, header fields are better IMO.
276 2017-07-17 11:57:03	0|jonasschnelli|Bitpays stuff is pretty nice (ECDSA auth via signature of post in header)
277 2017-07-17 11:57:07	0|luke-jr|couldn't do session-less for OAuth2's use case
278 2017-07-17 11:57:23	0|jonasschnelli|Just sign each request
279 2017-07-17 11:57:29	0|luke-jr|because the user is authenticating, and the app is just accessing that session
280 2017-07-17 11:57:42	0|jonasschnelli|And add an upcounting nonce
281 2017-07-17 11:57:59	0|luke-jr|jonasschnelli: the authenticator is not the API-accessor
282 2017-07-17 11:58:14	0|jonasschnelli|OAuth, yeah
283 2017-07-17 12:01:12	0|gmaxwell|the bitpay auth scheme is insecure. IIRC, has not anti-replay nonce
284 2017-07-17 12:02:09	0|gmaxwell|s/not/no/
285 2017-07-17 12:03:59	0|morcos|settxfee needs to go away
286 2017-07-17 12:04:18	0|morcos|we're getting close to being able to just do it on a per rpc call
287 2017-07-17 12:04:25	0|jonasschnelli|gmaxwell: Indeed
288 2017-07-17 12:04:35	0|luke-jr|morcos: probably replacing it with sessions isn't the way though :p
289 2017-07-17 13:06:07	0|wumpus|<morcos> settxfee needs to go away <- yes please
290 2017-07-17 13:06:29	0|bitcoin-git|13bitcoin/06master 14e061d8d 15practicalswift: Remove declaration of unused function: void UpdatedTransaction(const uint256 &)
291 2017-07-17 13:06:29	0|bitcoin-git|[13bitcoin] 15MarcoFalke pushed 2 new commits to 06master: 02https://github.com/bitcoin/bitcoin/compare/8bc6d1f179a0...2b0179d8a9b7
292 2017-07-17 13:06:30	0|bitcoin-git|13bitcoin/06master 142b0179d 15MarcoFalke: Merge #10834: Remove declaration of unused method: void UpdatedTransaction(const uint256 &)...
293 2017-07-17 13:06:59	0|bitcoin-git|[13bitcoin] 15MarcoFalke closed pull request #10834: Remove declaration of unused method: void UpdatedTransaction(const uint256 &) (06master...06remove-UpdatedTransaction) 02https://github.com/bitcoin/bitcoin/pull/10834
294 2017-07-17 13:07:35	0|wumpus|happy that we're moving to an API where all the important things can be set per transaction, so that different callers don't have to worry about messing up each other's global state
295 2017-07-17 13:11:21	0|promag|wumpus: remove https://github.com/bitcoin/bitcoin/pull/10650 from project 8?
296 2017-07-17 13:12:32	0|wumpus|only 3 things in project 8 anyhow - maybe we should clean it out for now, and use the 0.15 milestone list instead
297 2017-07-17 13:12:42	0|wumpus|(it's what I'm doing at least)
298 2017-07-17 13:13:03	0|wumpus|(one thing now)
299 2017-07-17 13:31:31	0|jonasschnelli|CodeShark: have you started to work on client side filtering after roasbeef's specs?
300 2017-07-17 13:42:12	0|jnewbery|Chris_Stewart_5: to get all blocks and transactions to propagate, try BitcoinTestFramework.sync_all()
301 2017-07-17 13:42:30	0|jnewbery|wumpus gmaxwell: for getinfo, "nag": "This RPC will be deprecated." will be silently dropped by any clients that are picking individual fields from getinfo.
302 2017-07-17 13:42:33	0|jnewbery|Hiding getinfo behind a flag for one version *forces* users to acknowledge that the RPC is being deprecated (and gives them one release grace period to fix their scripts/whatever that is using getinfo)
303 2017-07-17 13:42:57	0|jnewbery|unit testing or functional testing? Default keypool size in the functional tests is 1 (https://github.com/bitcoin/bitcoin/blob/master/test/functional/test_framework/test_framework.py#L212)
304 2017-07-17 13:42:57	0|jnewbery|wumpus: > keypool generation happens a lot during testing though
305 2017-07-17 13:44:43	0|wumpus|jnewbery: didn't know that the keypool size was overridden for testing, so in that case this doesn't help testing become faster, well too bad :)
306 2017-07-17 13:45:02	0|wumpus|jnewbery: yes, we can do the other thing for 0.16, just too late to make such a change for 0.15
307 2017-07-17 13:45:14	0|wumpus|jnewbery: I honestly don't understand why this comes up on the last day before feature freeze
308 2017-07-17 13:50:36	0|jnewbery|I agree, it's late. But hiding getinfo behind a flag in v0.16 => removing in 0.17. getinfo is one of the bitcoin_server -> bitcoin_wallet dependencies. I was hoping we'd be able to remove all of those for 0.16
309 2017-07-17 13:53:10	0|wumpus|sure, but couldn't you have brought it up a few weeks earlier?
310 2017-07-17 13:53:59	0|gmaxwell|I'd still complain that we haven't dogfooded it enough.
311 2017-07-17 13:55:17	0|wumpus|we could just remove the wallet balance from there
312 2017-07-17 13:55:30	0|gmaxwell|I wouldn't complain about that.
313 2017-07-17 13:55:33	0|wumpus|it makes no sense with multiwallet anyway
314 2017-07-17 13:56:20	0|wumpus|then again, that doesn't need to be done last-minute for 0.15, the circular dependency won't be solved before then anyhow
315 2017-07-17 13:56:43	0|wumpus|but let's say it's the last thing holding back solving the build dependency, we could do that
316 2017-07-17 13:57:19	0|wumpus|I still think adding the nag: makes sense, most people using getinfo will likely use it from the command line
317 2017-07-17 13:57:39	0|gmaxwell|even if they don't use it always that way, it's more visible than even release notes.
318 2017-07-17 13:57:50	0|wumpus|yes it would be in addition to release notes, not instead
319 2017-07-17 13:57:51	0|gmaxwell|and we've mentioned it in release notes before (I think, I hope)
320 2017-07-17 13:57:55	0|wumpus|yes, we did
321 2017-07-17 13:58:13	0|wumpus|even with the new places where everything could be found
322 2017-07-17 13:58:37	0|wumpus|https://github.com/bitcoin/bitcoin/blob/master/doc/release-notes/release-notes-0.14.0.md#getinfo-deprecated
323 2017-07-17 13:59:55	0|wumpus|I don't know why we waited for 0.14 before doing that
324 2017-07-17 14:04:29	0|BlueMatt|yeesh y'all been busy this morning
325 2017-07-17 14:04:38	0|wumpus|anyhow as it was deprecated in 0.14 (the deprecation message in the help was even backported to 0.13.1), removing it completely in 0.16 should be fine
326 2017-07-17 14:04:55	0|BlueMatt|wumpus: I mean its not a strong concern, really, I pointed it out cause it would be a problem for automated users (which I hope are mostly not calling -cli)
327 2017-07-17 14:05:52	0|wumpus|BlueMatt: yes, the -cli change would mostly be for command line users, but now agree we shouldn't do that, external stats/info scripts should be external
328 2017-07-17 14:06:27	0|BlueMatt|well i was more curious to find out what the demand for it was (re: the "IRC discussion" referenced in that pr)
329 2017-07-17 14:06:30	0|wumpus|and -cli should be a minimal, dumb client utiltiy
330 2017-07-17 14:06:38	0|gmaxwell|I think we'll have to do ~something~ but the best way to find that and get it done is just to live with it for a while.
331 2017-07-17 14:08:26	0|wumpus|maybe some example of a stats script? I don't know - generally, those things have gone unmaintained fast, as they're personal, everyone is interested in different info
332 2017-07-17 14:09:32	0|wumpus|there is also such a low threshold to writing them
333 2017-07-17 14:10:14	0|gmaxwell|connection count + height + errors is pretty universal; and is a bit obnoxious from the cli in our post getinfo world.
334 2017-07-17 14:11:08	0|jnewbery|> couldn't you have brought it up a few weeks earlier?
335 2017-07-17 14:11:22	0|wumpus|well for example personally I'm not interested in totai connection count - my script shows in/out versus ipv4/ipv6/tor
336 2017-07-17 14:11:49	0|jnewbery|I brought it up in response to #10841. I thought it would be friendlier to hide the RPC behind a flag before removing it one release later
337 2017-07-17 14:11:50	0|gribble|https://github.com/bitcoin/bitcoin/issues/10841 | [rpc] Give users one final warning before removing getinfo by jnewbery · Pull Request #10841 · bitcoin/bitcoin · GitHub
338 2017-07-17 14:12:08	0|wumpus|this is much more useful to troubleshoot any connectivity issues than the total number
339 2017-07-17 14:12:19	0|jnewbery|sorry, in response to #10838
340 2017-07-17 14:12:20	0|gribble|https://github.com/bitcoin/bitcoin/issues/10838 | (finally) remove the longest-ever-deprecated RPC call getinfo by TheBlueMatt · Pull Request #10838 · bitcoin/bitcoin · GitHub
341 2017-07-17 14:12:37	0|gmaxwell|wumpus: the normal thing I'm interested in is 0/8/many. seperate in/out counts would be interesting. From a cli perspective there are a number of other interesting things to show; e.g. best block hash, without getting into those huge info dumps like getblockchaininfo.
342 2017-07-17 14:12:56	0|wumpus|jnewbery: sure! BlueMatt was the original wtf 'why do you bring this up now', you tried to do a friendlier version, which is why I tagged that one
343 2017-07-17 14:12:58	0|jnewbery|If we're going to remove getinfo in v0.16, I suggest I close #10841
344 2017-07-17 14:12:59	0|gribble|https://github.com/bitcoin/bitcoin/issues/10841 | [rpc] Give users one final warning before removing getinfo by jnewbery · Pull Request #10841 · bitcoin/bitcoin · GitHub
345 2017-07-17 14:13:19	0|wumpus|jnewbery: but I stil think it is too much for such a late change
346 2017-07-17 14:13:49	0|wumpus|gmaxwell: exactly, none of which getinfo currently does, it's been frozen for years (on purpose)
347 2017-07-17 14:14:10	0|BlueMatt|i guess it doesnt matter much, though
348 2017-07-17 14:14:25	0|jnewbery|Yeah, doesn't matter too much. I'll close 10841.
349 2017-07-17 14:14:31	0|wumpus|we should have mentioned it in the release notes much earlier
350 2017-07-17 14:14:48	0|wumpus|jnewbery: I still think it makes sense to add a 'nag': to the result with your message, but anyhow
351 2017-07-17 14:14:49	0|gmaxwell|wumpus: I would have log ago proposed a 'status' rpc, but feared that the response would be "omg not another getinfo!"
352 2017-07-17 14:15:11	0|wumpus|gmaxwell: yes we're definitely not going to have any new server side call like that, that spans subsystems
353 2017-07-17 14:15:23	0|jnewbery|yeah, adding a "nag" field does no harm
354 2017-07-17 14:15:38	0|wumpus|client-side would have been ok with me, but runs into too much opposition, so we';re not going to do that either
355 2017-07-17 14:15:46	0|bitcoin-git|[13bitcoin] 15jnewbery closed pull request #10841: [rpc] Give users one final warning before removing getinfo (06master...06deprecate_getinfo) 02https://github.com/bitcoin/bitcoin/pull/10841
356 2017-07-17 14:16:48	0|gmaxwell|wumpus: if it didn't span into the wallet, but was just blockchain stats and p2p stats?
357 2017-07-17 14:17:34	0|wumpus|gmaxwell: I don't think that's a good idea, no
358 2017-07-17 14:17:42	0|gmaxwell|I figured.
359 2017-07-17 14:17:57	0|wumpus|could potentially create another dependency in the future that will be so hard to get rid of
360 2017-07-17 14:18:16	0|wumpus|if getting rid of getinfo would have been easier, I would have thought differently about it, but this keeps haunting us forever
361 2017-07-17 14:18:35	0|gmaxwell|well, we have to do something. I don't pretend to know what. But having to run multiple commands just to get the most basic of health information, is a real burden for support.
362 2017-07-17 14:18:59	0|BlueMatt|can we call #10526 a fix for a performance regression (really, disk space usage regression) and tag it 15 and merge it later this week?
363 2017-07-17 14:19:00	0|gribble|https://github.com/bitcoin/bitcoin/issues/10526 | Force on-the-fly compaction during pertxout upgrade by sipa · Pull Request #10526 · bitcoin/bitcoin · GitHub
364 2017-07-17 14:19:04	0|wumpus|then do something: write a useful script taht collects that information, submit it as PR
365 2017-07-17 14:19:06	0|wumpus|I TRIED
366 2017-07-17 14:19:19	0|gmaxwell|I think we haven't gotten rid of it because we haven't replaced it for anything but programmatic users who care little about how many calls they make or how many pages of uninteresting data they toss.
367 2017-07-17 14:19:31	0|wumpus|I just mistakingly made it part of -cli
368 2017-07-17 14:19:49	0|gmaxwell|wumpus: I know. Though I admit I have mixed feelings about cli being anything but a dumb client too. I am not yelling at you.
369 2017-07-17 14:19:57	0|gmaxwell|you did great.
370 2017-07-17 14:20:23	0|gmaxwell|BlueMatt: I believe we were waiting on feedback from wumpus to determine if there actually was a need there.
371 2017-07-17 14:20:39	0|wumpus|things that don't have to be done server side don't need to be done server side, this is the same reason I argue against #10804
372 2017-07-17 14:20:40	0|gribble|https://github.com/bitcoin/bitcoin/issues/10804 | Add histunspent RPC by promag · Pull Request #10804 · bitcoin/bitcoin · GitHub
373 2017-07-17 14:20:54	0|wumpus|client side has the information, so can roll the statistics and display them in any way
374 2017-07-17 14:21:02	0|gmaxwell|yes, I also saw 10804 and thought that.
375 2017-07-17 14:21:02	0|wumpus|the RPC interface doesn't have to acoomodate end users specifics
376 2017-07-17 14:21:28	0|gmaxwell|then our issue is that we just don't have a cli interface, except we have a highly used cli interface.
377 2017-07-17 14:21:31	0|gmaxwell|:)
378 2017-07-17 14:21:57	0|gmaxwell|BlueMatt: there was some question about the state of the node in question where he saw the not good results.
379 2017-07-17 14:22:00	0|wumpus|the same problem we had in 2012, and have had extensive discussions about since then, and no one is making that tool it seems
380 2017-07-17 14:22:14	0|wumpus|which I understand, I'm certainly not going to do it, too many opinions about it, I only write simple utility scripts for myself :)
381 2017-07-17 14:22:18	0|BlueMatt|wumpus: hmm, ok, now I understand the desire, thanks, yea thanks for making that pr. would be nice to just make it a bash script, but sadly fucking json :(
382 2017-07-17 14:23:11	0|wumpus|BlueMatt: it's trivial in python though
383 2017-07-17 14:23:21	0|BlueMatt|true
384 2017-07-17 14:23:21	0|gmaxwell|rename bitcoin-cli bitcoin-rpc and we create a real bitcoin-cli ? :P
385 2017-07-17 14:23:24	0|BlueMatt|lol
386 2017-07-17 14:23:39	0|mmgen|Just adding my two bits here: how about creating an addition cli command that aggregates info from various RPC calls and presents it in a readable way?
387 2017-07-17 14:23:41	0|wumpus|I have that script, could post it, but don't really feel like submitting such trivial things to the repo
388 2017-07-17 14:23:52	0|wumpus|mmgen: I DID THAT
389 2017-07-17 14:24:02	0|mmgen|wumpus: what's it called?
390 2017-07-17 14:24:03	0|wumpus|mmgen: no one liked it!
391 2017-07-17 14:24:08	0|BlueMatt|lol
392 2017-07-17 14:24:14	0|mmgen|I think that's the best solution
393 2017-07-17 14:24:22	0|mmgen|It could even be coded in Python
394 2017-07-17 14:24:35	0|wumpus|mmgen: I have a python and c++ implementation!
395 2017-07-17 14:25:18	0|gmaxwell|wumpus: I liked it except for one concern, the parity between the CLI and RPC is a major learing curve improvement. If we went down the path of a highly cooked bitcoin-cli then knoweldge wouldn't automatically transfer between them, improvements wouldn't automatically transfer, etc.
396 2017-07-17 14:25:20	0|wumpus|#8843 was the c++ one
397 2017-07-17 14:25:22	0|gribble|https://github.com/bitcoin/bitcoin/issues/8843 | rpc: Handle `getinfo` client-side in bitcoin-cli w/ `-getinfo` by laanwj · Pull Request #8843 · bitcoin/bitcoin · GitHub
398 2017-07-17 14:25:33	0|mmgen|wumpus: need to convince the others I think
399 2017-07-17 14:25:41	0|wumpus|I don't have the python one online, but could post it if anyone cares
400 2017-07-17 14:25:51	0|BlueMatt|yea, I'd be interested in putting the python one in contrib
401 2017-07-17 14:26:18	0|gmaxwell|for a getinfo replacement that cooked vs not cooked argument doesn't really hold, since it would just be one command.
402 2017-07-17 14:26:25	0|wumpus|gmaxwell: sure, that's just a matter of documentation though
403 2017-07-17 14:26:44	0|gmaxwell|which could be documented ('here are the underlying rpcs')
404 2017-07-17 14:26:45	0|wumpus|gmaxwell: it could show where it collects the various fields from for getinfo
405 2017-07-17 14:26:48	0|wumpus|yeah...
406 2017-07-17 14:26:53	0|mmgen|My wallet basically does just that, so I have a bit of experience in this area
407 2017-07-17 14:27:01	0|wumpus|a python script is self documentingi nthat regard
408 2017-07-17 14:27:10	0|wumpus|and easier to edit if you want something else
409 2017-07-17 14:27:23	0|wumpus|so ok, I'll add that to contrib, not in 0.15 though
410 2017-07-17 14:27:26	0|gmaxwell|wumpus: if it went as far as .e.g the ncurses interface, then that isn't the case. :)
411 2017-07-17 14:27:39	0|gmaxwell|wumpus: we could use more python rpc examples, IMO, in any case.
412 2017-07-17 14:28:06	0|wumpus|gmaxwell: yes I mean a simple getinfo replacement
413 2017-07-17 14:28:38	0|wumpus|ncurses isn't trivial (though also not rocket science, I've gotten people to start linux coding using it)
414 2017-07-17 14:29:18	0|promag|I'm happy to close #10804.. the problem is that to compute the histogram it takes really long to get and dump the json to the client with large utxo set
415 2017-07-17 14:29:23	0|gribble|https://github.com/bitcoin/bitcoin/issues/10804 | Add histunspent RPC by promag · Pull Request #10804 · bitcoin/bitcoin · GitHub
416 2017-07-17 14:29:36	0|mmgen|wumpus: ncurses is for wimps. real programmers use ANSI escape codes
417 2017-07-17 14:29:55	0|wumpus|mmgen: HAHA ansi escape codes are easier imo
418 2017-07-17 14:30:05	0|mmgen|wumpus: the fact is they are
419 2017-07-17 14:30:27	0|BlueMatt|morcos: wait, i thought min conftarget was 2?
420 2017-07-17 14:30:32	0|wumpus|(depending on what you want to do, if you need windowing and partial refresh etc, and menus, ncurses is definitely the way to go)
421 2017-07-17 14:31:04	0|mmgen|wumpus: true
422 2017-07-17 14:31:05	0|gmaxwell|promag: how long is really long? for how large?  With thousands of txouts it still takes a fraction of a second to listunspent last I checked.
423 2017-07-17 14:31:19	0|wumpus|if you want to go wild with displaying 24-bit color unicode graphs, ansi escape colors are definitely easier
424 2017-07-17 14:33:23	0|wumpus|promag: sure, computing the statistics server-side will be faster, but it's all a compromise...
425 2017-07-17 14:34:03	0|wumpus|promag: in that case maintaining a private patch set might make sense - the use for those statistics is probably specific to you
426 2017-07-17 14:34:27	0|gmaxwell|there are certantly things where I've written a python script to collect some data.. started it.... got tired of waiting, wrote it into the daemon on another host, recompiled, ran it, got the result...
427 2017-07-17 14:34:37	0|promag|gmaxwell: let me get fresh numbers for you
428 2017-07-17 14:34:59	0|gmaxwell|its unfortunate that rpc from python (and perhaps rpc in general) is so slow, though it is much faster than it used to be.
429 2017-07-17 14:35:27	0|BlueMatt|yea, Ive run python rpc clients that literally look days to complete
430 2017-07-17 14:35:29	0|mmgen|gmaxwell: I use it all the time. Haven't noticed that it's slow
431 2017-07-17 14:36:16	0|mmgen|but this is with my own RPC library
432 2017-07-17 14:36:30	0|wumpus|in that case it makes sense to implement it server-side in c++, but less sense for it to be merged upstream
433 2017-07-17 14:36:42	0|gmaxwell|mmgen: try doing something simple like graphing the size of all blocks from it... takes rather long time.  (or better, fee income from each block) at least with the normal python bitcoinrpc stuff
434 2017-07-17 14:37:24	0|wumpus|there's not really any API that can help you in that case, I guess there could be a dynlib plugin interface to load things into bitcoind, but does that ever make things easier than just recompiling?
435 2017-07-17 14:37:27	0|mmgen|gmaxwell: I don't use python-bitcoinrpc.  I just connect to the daemon directly
436 2017-07-17 14:38:01	0|gmaxwell|I believe we've factored up the relevant interfaces now that it shouldn't be hard to maintain external patches for things like this. Presumably we could take other changes that improved it further.
437 2017-07-17 14:38:50	0|gmaxwell|mmgen: may be that it's mostly the python side handling making it slow now. (didn't used to be the case, but the rpc has become a lot faster)
438 2017-07-17 14:39:12	0|mmgen|gmaxwell: Probably is. Python is slow
439 2017-07-17 14:39:18	0|mmgen|in general
440 2017-07-17 14:39:45	0|morcos|BlueMatt: huh?
441 2017-07-17 14:40:01	0|BlueMatt|morcos: new docs for estimatesmartfee says nblocks minimum is 1?
442 2017-07-17 14:40:11	0|BlueMatt|in #10707
443 2017-07-17 14:40:11	0|wumpus|sometimes speeding up python is easy, a lot of code works as-is with pypy
444 2017-07-17 14:40:13	0|gribble|https://github.com/bitcoin/bitcoin/issues/10707 | Better API for estimatesmartfee RPC by morcos · Pull Request #10707 · bitcoin/bitcoin · GitHub
445 2017-07-17 14:41:20	0|mmgen|wumpus: never tried pypy
446 2017-07-17 14:41:21	0|morcos|BlueMatt: you mean conf_target :)
447 2017-07-17 14:41:39	0|morcos|it is the minimum you can request, however asking for 1 will give you an answer for 2
448 2017-07-17 14:41:41	0|mmgen|wumpus: but everything remains single-threaded anyway
449 2017-07-17 14:41:44	0|BlueMatt|morcos: yes, sorry, was reviewing the first commit first :p
450 2017-07-17 14:41:52	0|BlueMatt|morcos: so lets just make the minimum 1?
451 2017-07-17 14:41:57	0|BlueMatt|(or at least mention that in the docs)
452 2017-07-17 14:42:05	0|morcos|it always worked to ask for 1, i didn't want to change that
453 2017-07-17 14:42:15	0|morcos|it seems like a semi-common thing a user might ask for
454 2017-07-17 14:42:17	0|BlueMatt|ok, so mention it in the docs, i guess
455 2017-07-17 14:42:23	0|wumpus|mmgen: it's still quite a big linear speedup for tight loops, better than you'd get with multithreading in most cases
456 2017-07-17 14:43:02	0|wumpus|(and it's no extra work, unlike multithreading)
457 2017-07-17 14:43:32	0|morcos|Also I think it preserves the ability to in the future return something different/smarter for a as fast as possible conf_target of 1
458 2017-07-17 14:47:59	0|morcos|BlueMatt: ok, while I'm at it, should I change the return parameter name from "blocks" now that we changed the argument to "conf_target" or leave it?
459 2017-07-17 14:48:18	0|morcos|I guess I vote leave it
460 2017-07-17 14:49:34	0|bitcoin-git|[13bitcoin] 15promag closed pull request #10804: Add histunspent RPC (06master...062017-07-rpc-add-histunspent) 02https://github.com/bitcoin/bitcoin/pull/10804
461 2017-07-17 14:49:48	0|BlueMatt|morcos: leave it, id say
462 2017-07-17 15:00:28	0|BlueMatt|morcos: "fee estimation is *only* able to return"....
463 2017-07-17 15:02:19	0|morcos|BlueMatt: did you read the whole sentence?
464 2017-07-17 15:02:38	0|morcos|that would make no sense
465 2017-07-17 15:03:22	0|BlueMatt|morcos: oh, its one sentence
466 2017-07-17 15:03:25	0|BlueMatt|ok, whatever
467 2017-07-17 15:04:36	0|morcos|just ACK it already
468 2017-07-17 15:05:02	0|BlueMatt|i did
469 2017-07-17 15:13:47	0|bitcoin-git|13bitcoin/06master 14dba485d 15Wladimir J. van der Laan: init: Factor out AppInitLockDataDirectory...
470 2017-07-17 15:13:47	0|bitcoin-git|[13bitcoin] 15laanwj pushed 2 new commits to 06master: 02https://github.com/bitcoin/bitcoin/compare/2b0179d8a9b7...89bb0365b97a
471 2017-07-17 15:13:48	0|bitcoin-git|13bitcoin/06master 1489bb036 15Wladimir J. van der Laan: Merge #10832: init: Factor out AppInitLockDataDirectory and fix startup core dump issue...
472 2017-07-17 15:14:17	0|bitcoin-git|[13bitcoin] 15laanwj closed pull request #10832: init: Factor out AppInitLockDataDirectory and fix startup core dump issue (06master...062017_07_appinitlockdatadirectory) 02https://github.com/bitcoin/bitcoin/pull/10832
473 2017-07-17 15:16:40	0|bitcoin-git|[13bitcoin] 15laanwj pushed 5 new commits to 06master: 02https://github.com/bitcoin/bitcoin/compare/89bb0365b97a...0b019357ff09
474 2017-07-17 15:16:41	0|bitcoin-git|13bitcoin/06master 1430d8f3a 15Gregory Maxwell: Pushdown walletdb though CWallet::AddKeyPubKey to avoid flushes....
475 2017-07-17 15:16:41	0|bitcoin-git|13bitcoin/06master 143a53f19 15Gregory Maxwell: Pushdown walletdb object through GenerateNewKey/DeriveNewChildKey....
476 2017-07-17 15:16:42	0|bitcoin-git|13bitcoin/06master 1441dc163 15Gregory Maxwell: Increase wallet default keypool size to 1000.
477 2017-07-17 15:17:12	0|bitcoin-git|[13bitcoin] 15laanwj closed pull request #10831: Batch flushing operations to the walletdb during top up and increase keypool size. (06master...06topup_batch_flush) 02https://github.com/bitcoin/bitcoin/pull/10831
478 2017-07-17 15:19:48	0|cfields|wumpus: re qt, no, i don't think it's worth messing with for now
479 2017-07-17 15:22:08	0|wumpus|ok
480 2017-07-17 15:23:52	0|BlueMatt|gah, github is missing notification emails :(
481 2017-07-17 15:26:46	0|BlueMatt|wumpus: hmmm....I think (in addition to github actually missing notification emails) github wont send an email for a merge unless you comment merged.
482 2017-07-17 15:26:49	0|BlueMatt|can you do that?
483 2017-07-17 15:27:06	0|wumpus|hm?
484 2017-07-17 15:27:12	0|morcos|BlueMatt: I've never gotten more than about 1 in 3 of my expected github notification emails
485 2017-07-17 15:27:14	0|BlueMatt|eg #10831
486 2017-07-17 15:27:16	0|gribble|https://github.com/bitcoin/bitcoin/issues/10831 | Batch flushing operations to the walletdb during top up and increase keypool size. by gmaxwell · Pull Request #10831 · bitcoin/bitcoin · GitHub
487 2017-07-17 15:27:21	0|BlueMatt|morcos: did you mail support@?
488 2017-07-17 15:27:24	0|wumpus|github should send an automated mail when something gets merged
489 2017-07-17 15:27:30	0|BlueMatt|morcos: i just for the first time missed some emails
490 2017-07-17 15:27:32	0|morcos|i complained to sdaftuar, does that count
491 2017-07-17 15:27:36	0|BlueMatt|........
492 2017-07-17 15:27:45	0|morcos|it didn't used to happen to him, but now it does too
493 2017-07-17 15:27:52	0|BlueMatt|well whatever, I have all mailserver logs and didnt get any, so I'm gonna bitch to support@
494 2017-07-17 15:27:52	0|wumpus|but I don't know what you mean with 'comment merged'
495 2017-07-17 15:28:06	0|BlueMatt|wumpus: as in usually when sipa merges he comments on the pr merged when he closes it
496 2017-07-17 15:28:07	0|BlueMatt|iirc
497 2017-07-17 15:28:13	0|BlueMatt|saying "Merged"
498 2017-07-17 15:28:21	0|wumpus|huh? no, he doesn't
499 2017-07-17 15:29:18	0|BlueMatt|oh?
500 2017-07-17 15:29:24	0|wumpus|see https://github.com/bitcoin/bitcoin/pull/10735 https://github.com/bitcoin/bitcoin/pull/10844
501 2017-07-17 15:29:36	0|wumpus|https://github.com/bitcoin/bitcoin/pull/10840
502 2017-07-17 15:30:00	0|BlueMatt|hmm, oh, you're right!
503 2017-07-17 15:30:04	0|wumpus|being the last things he merged, he commented on none of them, besides an utACK in some cases, but no "Merged". Github does that.
504 2017-07-17 15:30:07	0|BlueMatt|ugh, yea, so github notifications ar ejust fucked
505 2017-07-17 15:30:15	0|BlueMatt|yea, ok, ill go bitch at them
506 2017-07-17 15:31:32	0|BlueMatt|clearly we should be doing dev on our own hosted infrastructure so that we dont miss notification emails :p
507 2017-07-17 15:32:01	0|wumpus|as if sending mails from your own infrastructure (and having them actually arrive) is a sure thing, or was that the joke? :)
508 2017-07-17 15:32:09	0|BlueMatt|it was a joke
509 2017-07-17 15:32:10	0|BlueMatt|yea
510 2017-07-17 15:32:10	0|gmaxwell|"now you have two problems"
511 2017-07-17 15:32:20	0|BlueMatt|i mean they'd arrive for me, cause i run the receiving mailserver too :p
512 2017-07-17 15:33:04	0|gmaxwell|BlueMatt: but feel free to setup mirrored infra... :)
513 2017-07-17 15:33:12	0|BlueMatt|lol, I'll pass
514 2017-07-17 15:33:16	0|BlueMatt|i already manage too much shit
515 2017-07-17 15:38:11	0|gmaxwell|What do we think of #10817 ?  I didn't notice until a day ago that it wasn't 15 flagged, I think it's ready to go.  And its a likely useful lead up to work in 0.16-- in that it will increase change discarding somewhat, while planned work for 0.16 (the exact match coin selecter) will increase it a lot more... might be good to have in field feedback (are people going to be irate about it throwing
516 2017-07-17 15:38:12	0|gribble|https://github.com/bitcoin/bitcoin/issues/10817 | Redefine Dust and add a discard_rate by morcos · Pull Request #10817 · bitcoin/bitcoin · GitHub
517 2017-07-17 15:38:17	0|gmaxwell|away an extra 5 cents in bitcoin to avoid change outputs)
518 2017-07-17 15:39:09	0|wumpus|tagged
519 2017-07-17 15:39:21	0|gmaxwell|danke.
520 2017-07-17 15:39:43	0|bitcoin-git|[13bitcoin] 15theuni opened pull request #10851: depends: bump fontconfig to 2.12.4 (06master...06fontconfig-bump) 02https://github.com/bitcoin/bitcoin/pull/10851
521 2017-07-17 15:48:45	0|gmaxwell|I think 9728 can be untagged for 15, retagged for 16.
522 2017-07-17 15:49:37	0|instagibbs|freeze is after today yes?
523 2017-07-17 15:49:50	0|wumpus|gmaxwell: ok, agree
524 2017-07-17 15:49:53	0|instagibbs|ack re 9728, we still have work on it
525 2017-07-17 15:50:02	0|instagibbs|but should def be ready for 16
526 2017-07-17 15:50:03	0|wumpus|instagibbs: seems we're slipping with regard to multiwallet
527 2017-07-17 15:50:24	0|instagibbs|yeah I noticed, but I don't have the strongest opinion so kinda ducked out of that
528 2017-07-17 15:50:28	0|wumpus|instagibbs: but yeah, please no new stuff now
529 2017-07-17 15:51:02	0|wumpus|(unless they're bugfixes, ofcourse)
530 2017-07-17 15:51:17	0|gmaxwell|that was the only thing in the 15 tags that seemed to me that obviously wouldn't make it. (I'm sure some other things won't too)
531 2017-07-17 15:53:10	0|morcos|#10830 is missing the 0.15 tag and probably still needs the most work
532 2017-07-17 15:53:12	0|gribble|https://github.com/bitcoin/bitcoin/issues/10830 | [WIP] [wallet] keypool restore by jnewbery · Pull Request #10830 · bitcoin/bitcoin · GitHub
533 2017-07-17 15:53:14	0|BlueMatt|ok, what are we doing about multiwallet for 15?
534 2017-07-17 15:53:36	0|gmaxwell|morcos: Agreed, though I consider 10830 a bugfix. It should still be tagged. (the other PR was tagged)
535 2017-07-17 15:54:10	0|gmaxwell|sipa: Have you looked at the redo of the endpoint change... it's much simpler and drop-in usable than the earlier one.
536 2017-07-17 15:55:17	0|bitcoin-git|[13bitcoin] 15jnewbery opened pull request #10853: [tests] Fix RPC failure testing (again) (06master...06cleanup_jsonrpc_asserts) 02https://github.com/bitcoin/bitcoin/pull/10853
537 2017-07-17 15:55:28	0|morcos|As for multiwallet, everyone just vote 0, 1, or 2 votes for either #10829 or #10849 based on how stronly you feel, and lets just pick one. honestly, we'll be fine with either choice, but lets concentrate our effort on one of them
538 2017-07-17 15:55:30	0|gribble|https://github.com/bitcoin/bitcoin/issues/10829 | Simple, backwards compatible RPC multiwallet support. by ryanofsky · Pull Request #10829 · bitcoin/bitcoin · GitHub
539 2017-07-17 15:55:31	0|gribble|https://github.com/bitcoin/bitcoin/issues/10849 | Multiwallet: simplest endpoint support by jonasschnelli · Pull Request #10849 · bitcoin/bitcoin · GitHub
540 2017-07-17 15:55:51	0|gmaxwell|#10821 tag for 0.16
541 2017-07-17 15:55:52	0|gribble|https://github.com/bitcoin/bitcoin/issues/10821 | Add SSE4 optimized SHA256 by sipa · Pull Request #10821 · bitcoin/bitcoin · GitHub
542 2017-07-17 15:56:06	0|morcos|I vote 0 votes
543 2017-07-17 15:56:25	0|gmaxwell|do not vote for SSE4 optimized multiwallet.
544 2017-07-17 15:56:38	0|BlueMatt|oooo, 2 votes for that (whatever the hell it is)
545 2017-07-17 15:56:57	0|morcos|as my 2 year old would say whenever we tell him no..  "not yet?"
546 2017-07-17 15:56:58	0|instagibbs|1.6x as secure as regular multiwallet
547 2017-07-17 16:02:38	0|jnewbery|Is #10650 definitely dead? It had 4 ACKs and instagibbs and I had both tested it.
548 2017-07-17 16:02:41	0|gribble|https://github.com/bitcoin/bitcoin/issues/10650 | Multiwallet: add RPC endpoint support by jonasschnelli · Pull Request #10650 · bitcoin/bitcoin · GitHub
549 2017-07-17 16:03:12	0|jnewbery|If it is dead, then my 1 vote is for #10849, since I've reviewed and tested that manually, and I think endpoints is the correct way to go
550 2017-07-17 16:03:14	0|gribble|https://github.com/bitcoin/bitcoin/issues/10849 | Multiwallet: simplest endpoint support by jonasschnelli · Pull Request #10849 · bitcoin/bitcoin · GitHub
551 2017-07-17 16:03:21	0|instagibbs|1 very vehement NACK, basically
552 2017-07-17 16:03:27	0|instagibbs|(not saying right or wrong)
553 2017-07-17 16:04:15	0|jnewbery|but if other people have reviewed and tested #10829 then I can also live with that. I see the multiwallet interface in v0.15 as unstable/experimental, so as long as something gets in, I'm happy
554 2017-07-17 16:04:17	0|gribble|https://github.com/bitcoin/bitcoin/issues/10829 | Simple, backwards compatible RPC multiwallet support. by ryanofsky · Pull Request #10829 · bitcoin/bitcoin · GitHub
555 2017-07-17 16:05:07	0|instagibbs|I mostly reviewed that things behaved like they should, I can do the same today for either of the other two. I'm happy with whatever works.
556 2017-07-17 16:07:00	0|gmaxwell|-1 on 10650, I think  like 10849 better than 10829, as I think it will be more usable in practice but I will be fully willing to help throw out either interface later, so I don't care overly much.  10849 is a newer pr... both presumably still need a lot of review love.
557 2017-07-17 16:12:19	0|gmaxwell|morcos: care to review #10672
558 2017-07-17 16:12:20	0|gribble|https://github.com/bitcoin/bitcoin/issues/10672 | Avoid division by zero in the case of a corrupt estimates file by practicalswift · Pull Request #10672 · bitcoin/bitcoin · GitHub
559 2017-07-17 16:17:02	0|cfields|BlueMatt: are you still wanting #10652 for 0.15 ?
560 2017-07-17 16:17:04	0|gribble|https://github.com/bitcoin/bitcoin/issues/10652 | Small step towards demangling cs_main from CNodeState by TheBlueMatt · Pull Request #10652 · bitcoin/bitcoin · GitHub
561 2017-07-17 16:20:06	0|gmaxwell|#10335 sounds like something should be done.
562 2017-07-17 16:20:08	0|gribble|https://github.com/bitcoin/bitcoin/issues/10335 | back-compat: add fallback getentropy implementation by theuni · Pull Request #10335 · bitcoin/bitcoin · GitHub
563 2017-07-17 16:35:06	0|cfields|gmaxwell: thanks for the reminder. fixing that up now.
564 2017-07-17 16:41:30	0|BlueMatt|fizzwont: nah, think its gonna slip now
565 2017-07-17 16:41:33	0|BlueMatt|cfields: ...
566 2017-07-17 16:41:44	0|instagibbs|little worried about 10829 since named support is still a bit flakey. I think merging that would mean fixing up named arg support becomes priority.
567 2017-07-17 16:41:49	0|BlueMatt|it would've been nice, but i dont think its critical enough to care, just untag 10672
568 2017-07-17 16:42:20	0|gmaxwell|BlueMatt: whats the 'fizzwont' context for your message?
569 2017-07-17 16:42:36	0|BlueMatt|gmaxwell: bad auto-tab, its a user on here
570 2017-07-17 16:42:40	0|cfields|:(
571 2017-07-17 16:43:12	0|BlueMatt|cfields: meh, its mostly a bugfix for not-really-issues bugs
572 2017-07-17 16:43:21	0|BlueMatt|the cleanup to do the demangle is gonna be a through-16 process
573 2017-07-17 16:44:18	0|fizzwont|i'm honored...but just observing
574 2017-07-17 17:03:26	0|bitcoin-git|[13bitcoin] 15gmaxwell opened pull request #10854: Avoid using sizes on non-fixed-width types to derive protocol constants. (06master...06rbf-numlimit-fix) 02https://github.com/bitcoin/bitcoin/pull/10854
575 2017-07-17 17:15:58	0|bitcoin-git|[13bitcoin] 15theuni closed pull request #10335: back-compat: add fallback getentropy implementation (06master...06getentropy-back-compat) 02https://github.com/bitcoin/bitcoin/pull/10335
576 2017-07-17 17:18:08	0|bitcoin-git|[13bitcoin] 15theuni opened pull request #10855: random: only use getentropy on openbsd (06master...06getentropy-openbsd) 02https://github.com/bitcoin/bitcoin/pull/10855
577 2017-07-17 17:52:11	0|morcos|yeah someone please tag #10758 for 0.15
578 2017-07-17 17:53:28	0|sipa|done
579 2017-07-17 17:54:21	0|morcos|instagibbs: to be clear, there are no rpc calls where you can't use named arguments right?  there just might be some where you need to specifically specify the default value if you want to name a later argument (but this is the same thing you have to do with positional)
580 2017-07-17 17:54:55	0|sipa|wasnt't there a PR to fix that?
581 2017-07-17 17:55:19	0|instagibbs|morcos, oh, true
582 2017-07-17 17:55:29	0|instagibbs|yes you'll have to enter in all values for some
583 2017-07-17 17:55:37	0|gribble|https://github.com/bitcoin/bitcoin/issues/10758 | Fix some chainstate-init-order bugs. by TheBlueMatt · Pull Request #10758 · bitcoin/bitcoin · GitHub
584 2017-07-17 17:55:45	0|instagibbs|so, less urgent :)
585 2017-07-17 17:55:48	0|morcos|sipa: yeah, i was just trying to understand the urgency
586 2017-07-17 17:56:03	0|instagibbs|merely annoying
587 2017-07-17 17:57:26	0|BlueMatt|lol, somehow i thought 10758 was already tagged 15
588 2017-07-17 17:57:29	0|morcos|i guess people didn't like my voting idea or don't have an opinon on which multiwallet PR to merge.  can we decide so we can make it happen?  i'm happy to review 10849 if we prefer that one
589 2017-07-17 17:57:48	0|sipa|i'll review the last PR before giving an opinion
590 2017-07-17 18:03:28	0|ryanofsky|i'd give one vote to 10829 (obvs), and one conditional vote to 10849 if it uses a plain query parameter, or has some documentation explaining the path schema
591 2017-07-17 18:05:42	0|sipa|i think a query parameter is inferior to just a named parameter, and doesn't really simplify later moving to a new process
592 2017-07-17 18:06:10	0|sipa|i'm fine with named parameters for now, as a stopgap measure because a full new API isn't complete
593 2017-07-17 18:06:16	0|sipa|documentation can come later
594 2017-07-17 18:06:27	0|sipa|that doesn't need to be ready by the feature freeze
595 2017-07-17 18:08:03	0|sipa|but i do think a new endpoint is eventually the way to go - whether that happens in 0.15 or not
596 2017-07-17 18:08:06	0|ryanofsky|sipa i think moving to new process is basically orthogonal. reason why i think query parameter is better is it leave uri-path space open for future uses and an actual thoughtful design. current design is changing multiple times a day
597 2017-07-17 18:08:48	0|ryanofsky|add v1, subtract v1, forward node calls through wallet, then stop, then start again
598 2017-07-17 18:09:17	0|sipa|ryanofsky: i really disagree with that... the point of an endpoint is that clients can be configured to point to a new process
599 2017-07-17 18:09:37	0|ryanofsky|i'm trying to understand...
600 2017-07-17 18:09:41	0|sipa|you can't do that with query parameters
601 2017-07-17 18:09:56	0|ryanofsky|? why not?
602 2017-07-17 18:10:20	0|sipa|what if some of them are process-specific, and some or not?
603 2017-07-17 18:10:32	0|sipa|the client software would need to go modify the url to put in extra things
604 2017-07-17 18:11:00	0|sipa|i guess it isn't worse - as long as everything you'd put in the path becomes a query parameter and nothing else
605 2017-07-17 18:11:05	0|sipa|but it also doesn't gain you anything
606 2017-07-17 18:11:06	0|jtimon|ryanofsky: it's about simplyfing concurrency
607 2017-07-17 18:11:11	0|sipa|jtimon: wut?
608 2017-07-17 18:11:18	0|jtimon|is it not?
609 2017-07-17 18:11:24	0|sipa|what does that have to do with anything
610 2017-07-17 18:11:59	0|ryanofsky|i'm not understanding what you're saying sipa... did you change your mind midway?
611 2017-07-17 18:12:29	0|sipa|ryanofsky: yes, i did; i agree that query parameters can do anything that a URI can... but it also doesn't gain you anything
612 2017-07-17 18:12:33	0|ryanofsky|advantage of query parameter is it leaves more options for the future that don't require us to break compatibility
613 2017-07-17 18:12:40	0|ryanofsky|nothing to do with multiprocess
614 2017-07-17 18:13:54	0|ryanofsky|specific example of what i'm talking about: maybe we want to have web interface listening on /wallet. maybe we want to add versioning in the future
615 2017-07-17 18:14:00	0|sipa|ryanofsky: but the point is creating a namespace
616 2017-07-17 18:14:13	0|jtimon|right, each wallet can get its own lock even with query paramters, then I don't undesrtand what's the point either...can't you serve the rpc more concurrently with different http addresses more concurrently either (ie maybe a server for each or something)?
617 2017-07-17 18:14:15	0|sipa|separating things off, which can later move elsewhere
618 2017-07-17 18:14:24	0|ryanofsky|doing these things cleanly means planning out url design somewhat, which hasn't been done. in these prs url design is constantly in flux
619 2017-07-17 18:14:34	0|sipa|jtimon: that's completely orthogonal; we're talking about interface here
620 2017-07-17 18:15:00	0|jtimon|well, I remembered that about the most convincing argument in favor of this interface
621 2017-07-17 18:15:14	0|jtimon|s/about/as/
622 2017-07-17 18:15:22	0|sipa|jtimon: we already have multithreaded RPC...
623 2017-07-17 18:15:38	0|jtimon|sipa: oh, nice
624 2017-07-17 18:15:41	0|sipa|jtimon: ...
625 2017-07-17 18:16:26	0|ryanofsky|jtimon, fwiw, i don't see connection there either
626 2017-07-17 18:16:51	0|sipa|jtimon: the RPC implementation isn't very concurrent, but that's an implementation detail that can be improved independently
627 2017-07-17 18:17:01	0|jtimon|I didn't know, sorry, then I don't understand why this wallet/<my wallet> is better than wallet=<mywallet> in the rpc either. but I'll read your conversation and see if I get it
628 2017-07-17 18:17:01	0|sipa|(overeager locking in many places)
629 2017-07-17 18:17:20	0|sipa|jtimon: it's so that things can move to a different process
630 2017-07-17 18:17:41	0|sipa|if the wallet weren't handled by bitcoind anymore, it would have its own RPC server, which runs on a different port or something
631 2017-07-17 18:17:54	0|jtimon|sipa: how is moving things to a different process unrelated to concurrency?
632 2017-07-17 18:18:04	0|sipa|jtimon: because it's already multithreaded!
633 2017-07-17 18:18:27	0|jtimon|then what's the point of moving it to a different process?
634 2017-07-17 18:18:38	0|sipa|the reason for moving things to a different process is for security (don't have your private keys connected to a network interface) and modularity (run it on a different machine)
635 2017-07-17 18:19:05	0|jtimon|ok, thanks, I wrongly assumed the reason was concurrency
636 2017-07-17 18:19:07	0|ryanofsky|my the way my conditional vote for 10849 had an OR in it. if query parameter is worse for some reason (aesthetics?) and path interpretation is way to go, please just document/respond what actual plans are for path interpretation. what compatibility is being promised
637 2017-07-17 18:19:24	0|sipa|ryanofsky: of course it needs documenting
638 2017-07-17 18:20:16	0|sipa|ryanofsky: if we aren't clear on how the namespace separation should happen, i agree that just an RPC named parameter for selecting a wallet is better for now
639 2017-07-17 18:20:39	0|sipa|ryanofsky: but if we are, we should do it... and url parameters seem like a very odd thing for a namespace
640 2017-07-17 18:21:39	0|ryanofsky|i don't understand the point about a namespace. is an aesthetic thing, or a practical difference?
641 2017-07-17 18:22:30	0|ryanofsky|i'm asking for documentation not just because documentation is needed, but also because i really think i am missing something about rationales (or that rationales haven't actually been thought through)
642 2017-07-17 18:22:46	0|sipa|ryanofsky: my concern is that uri parameters don't force us to think about separation
643 2017-07-17 18:22:58	0|jtimon|but the parameters don't need to go in the url, do they? they can go with the json data (ie with method, params, jsonrpc and id)
644 2017-07-17 18:23:19	0|sipa|jtimon: yes, that's the discussion... there are 4 approaches suggested
645 2017-07-17 18:24:12	0|jtimon|I think I'm missing one from the last meeting...
646 2017-07-17 18:24:55	0|sipa|jtimon: in the URL (http://localhost:8333/v1/wallet/[walletname]), in a URL parameter (http://localhost:8333/v1?wallet=[walletname]), in a JSON RPC parameter (pass wallet: "[walletname]" as named argument), through RPC auth (every rpc user has his own wallet)
647 2017-07-17 18:25:25	0|jtimon|I see, rpc auth was the one I was mising
648 2017-07-17 18:25:45	0|sipa|ryanofsky: people may be inclined to add things to URI parameters that don't naturally correspond to separation (say, an option for changing the default currency unit)
649 2017-07-17 18:25:52	0|sipa|ryanofsky: or expect those to go there
650 2017-07-17 18:25:56	0|jtimon|no, I thought we agreed some users will have more than one, and I think some users may not want one too
651 2017-07-17 18:26:32	0|sipa|ryanofsky: sure you can say, "well don't do that"
652 2017-07-17 18:26:45	0|ryanofsky|not sure why a query option like that would be bad at all. do you also think a timeout query option would be bad?
653 2017-07-17 18:26:46	0|BlueMatt|it may be unfair, but ryanofsky succcessfully convinced me over lunch....the user experience of upgrading is gonna be the same no matter how we do it, and the args approach is much simpler for users to add
654 2017-07-17 18:26:53	0|BlueMatt|rather than the endpoints approach
655 2017-07-17 18:27:23	0|sipa|ryanofsky: ok, say we have ?wallet=[walletname]&timeout=30
656 2017-07-17 18:27:32	0|sipa|all in one process
657 2017-07-17 18:27:41	0|BlueMatt|the endpoints stuff is kinda nice in that it makes users thing, but realistically its just another hop to upgrade, cause if we move to process isolation now its all a different ip/port connection anyway
658 2017-07-17 18:27:47	0|BlueMatt|so its just a useless upgrade hop
659 2017-07-17 18:28:13	0|sipa|now the wallets move to different processes
660 2017-07-17 18:28:31	0|BlueMatt|and now the split is by connection port, and not endpoint anyway :p
661 2017-07-17 18:28:40	0|sipa|you can't just go reconfigure the client software
662 2017-07-17 18:29:08	0|sipa|it needs to mix something from the configured URI (which likely has the ?wallet= part) and from the application (which likely has the ?timeout= part)
663 2017-07-17 18:29:10	0|BlueMatt|actually args makes it easier at that point than endpoints?
664 2017-07-17 18:29:16	0|BlueMatt|precisely cause it doesnt require client reconfiguration
665 2017-07-17 18:29:29	0|BlueMatt|wait, now maybe I'm confused, why are we re-adding uri parameters now
666 2017-07-17 18:29:34	0|sipa|but client reconfiguration is easy
667 2017-07-17 18:29:36	0|ryanofsky|sipa are you talking about a node process that forwards requests to various wallets?
668 2017-07-17 18:29:41	0|sipa|ryanofsky: no
669 2017-07-17 18:29:42	0|ryanofsky|BlueMatt, i have same confusion
670 2017-07-17 18:30:02	0|ryanofsky|so you are talking about what matt is talking about. wallet process listening on own port
671 2017-07-17 18:30:08	0|BlueMatt|doing wallet split today is very doable - have multiple rpc clients pointed at different listening ports
672 2017-07-17 18:30:10	0|sipa|i want to just tell my client app (instead of http://localhost:8333, you can use http://localhost:8333/v1/wallet/bla)
673 2017-07-17 18:30:22	0|BlueMatt|?
674 2017-07-17 18:30:22	0|BlueMatt|but you cant with most clients/
675 2017-07-17 18:30:31	0|ryanofsky|wait sipa, that sounds like you are saying one port for node and wallet?
676 2017-07-17 18:30:39	0|sipa|ryanofsky: ???
677 2017-07-17 18:30:55	0|sipa|as long as they're in the same process
678 2017-07-17 18:31:01	0|ryanofsky|port 8333 is wallet process or node process?
679 2017-07-17 18:31:07	0|sipa|it's the process process
680 2017-07-17 18:31:09	0|sipa|there is just one
681 2017-07-17 18:31:10	0|BlueMatt|instead of btc = AuthServiceProxy(); btc.getwalletinfo(); btc.getmininginfo(); its now wallet_btc = AuthServiceProxy(...:8332); node_btc = AuthServiceProxy(...:8331)......
682 2017-07-17 18:31:23	0|sipa|ryanofsky: i'm talking about the case before things move to a idfferent process
683 2017-07-17 18:31:25	0|BlueMatt|sipa: I'm horribly confused
684 2017-07-17 18:31:28	0|ryanofsky|AH ok
685 2017-07-17 18:31:34	0|ryanofsky|so single process just like today
686 2017-07-17 18:32:13	0|BlueMatt|sipa: I think ryanofsky's point (and my point) is that we should focus on making it easier for clients to do the upgrade (which wallet arg is, imo) cause the isolation is gonna come from different listening ports in the future anyway
687 2017-07-17 18:32:15	0|sipa|however, afterwards, when things move to a new process, you don't need to change the application software, just reconfigure it to use a different URI (which is now maybe running on a different port, or a different host)
688 2017-07-17 18:32:25	0|BlueMatt|so trying to force endpoints is just an extra step for users for no reason
689 2017-07-17 18:32:41	0|BlueMatt|sipa: you can already do that?
690 2017-07-17 18:32:48	0|ryanofsky|sipa, how is wallet named arg a problem in that scenario?
691 2017-07-17 18:32:54	0|BlueMatt|just tell your app software that you have two different rpc hosts
692 2017-07-17 18:33:00	0|ryanofsky|i think wallet named arg is a feature even in that scenario
693 2017-07-17 18:33:15	0|sipa|ryanofsky: because wallet named arg requires at the very least the client application to know which wallet it is talking to
694 2017-07-17 18:33:23	0|ryanofsky|because it's easy to imagine starting wallet processes and confusing yourself about port numbers
695 2017-07-17 18:33:30	0|sipa|while it can be encapsulated in the uri
696 2017-07-17 18:34:52	0|BlueMatt|sipa: if you're gonna change the rpc library to support endpoints, you can also change it to support a global wallet argument
697 2017-07-17 18:35:01	0|ryanofsky|i'm confused again. now we are talking about separate wallet process listening it's separate wallet port. what is the harm of wallet param there (used or unused)?
698 2017-07-17 18:35:13	0|BlueMatt|whereas arguments are nice cause you dont have to change the library, if it already supports named args
699 2017-07-17 18:35:14	0|sipa|BlueMatt: but after process separation there may not even be a need for a global wallet argument
700 2017-07-17 18:35:23	0|jtimon|for what is worth, I like v1 in the uri more than "jsonrpc": "1.0" in the json data, but do we need 2 of them?
701 2017-07-17 18:35:24	0|sipa|BlueMatt: endpoint support means there is just one change
702 2017-07-17 18:35:33	0|BlueMatt|sipa: sure, and its, in fact, easier to remove the wallet arg at that point than the endpoints
703 2017-07-17 18:35:34	0|sipa|to the application
704 2017-07-17 18:35:40	0|BlueMatt|cause we can even start ignoring the arg
705 2017-07-17 18:36:11	0|sipa|you really think so? that first requiring everything to add a wallet configuration option, and then later change it again because now it's done in a different URI
706 2017-07-17 18:36:17	0|sipa|is easier than just allowing to configure a URI?
707 2017-07-17 18:36:44	0|ryanofsky|oh sipa, now i finally get it, yes i agree
708 2017-07-17 18:36:51	0|BlueMatt|sipa: think about eg the python library which is never updated
709 2017-07-17 18:36:59	0|BlueMatt|you could use multiwallet even before the library is updated
710 2017-07-17 18:37:08	0|BlueMatt|and if you update the library then its the same
711 2017-07-17 18:37:18	0|ryanofsky|yes that is an argument against named json-rpc params in favor of urls
712 2017-07-17 18:37:20	0|BlueMatt|(either the library silently adds the wallet arg to all calls, or it doesnt)
713 2017-07-17 18:37:22	0|sipa|i'm confused
714 2017-07-17 18:37:29	0|BlueMatt|instead of silently adding the wallet endpoint to all calls, or not
715 2017-07-17 18:37:40	0|ryanofsky|it's not an argument for wallet in uri-path instead of wallet in url query-string
716 2017-07-17 18:37:44	0|sipa|BlueMatt: it shouldn't silently add wallet endpoints!
717 2017-07-17 18:37:54	0|sipa|BlueMatt: you should tell it "MY URL IS ...../v1/wallet"
718 2017-07-17 18:38:08	0|sipa|it shouldn't even know there is such a thing as wallet names
719 2017-07-17 18:38:09	0|BlueMatt|sipa: does the current python client support that?
720 2017-07-17 18:38:19	0|BlueMatt|(I dont know, I'm asking)
721 2017-07-17 18:38:20	0|sipa|BlueMatt: i don't know, but it'd be trivial to add
722 2017-07-17 18:38:38	0|sipa|and certainly easier than updating it to selectively go add wallet parameters everywhere
723 2017-07-17 18:38:42	0|sipa|and then later making that optional
724 2017-07-17 18:39:29	0|jtimon|sipa: how is "encapsulated in the uri" different from "encapsulated on the json data like 'jsonrpc' and 'method'"? if that's better, why not move the method to the uri too?
725 2017-07-17 18:39:51	0|sipa|jtimon: because thr rpc method and parameter are decided by the application
726 2017-07-17 18:40:00	0|sipa|jtimon: and the URI is decided by the person configuring the software
727 2017-07-17 18:40:03	0|jtimon|sipa: and so it's the uri?
728 2017-07-17 18:40:09	0|sipa|no
729 2017-07-17 18:40:21	0|sipa|you tell your application which host and port the RPC server runs on,
730 2017-07-17 18:40:26	0|sipa|you can also tell it what URI to use
731 2017-07-17 18:40:46	0|sipa|ryanofsky: i agree that technically url query-string can do as much as uri-path... but i do believe that this approach can only work if we clearly think about what can be separated
732 2017-07-17 18:40:47	0|BlueMatt|sipa: issue is I do not believe most client libraries support that today
733 2017-07-17 18:40:49	0|jtimon|mhmm, well, that's were I'm confused, I don't see how you could have control of the json data bur not of the uri
734 2017-07-17 18:40:52	0|sipa|BlueMatt: sure, so?
735 2017-07-17 18:40:54	0|BlueMatt|(I was informed that the python-bitcoinrpc library does not)
736 2017-07-17 18:40:55	0|jtimon|s/were/where
737 2017-07-17 18:40:55	0|sipa|BlueMatt: they can be updated
738 2017-07-17 18:41:16	0|BlueMatt|sipa: whereas client libraies may already support a wallet argument
739 2017-07-17 18:41:18	0|BlueMatt|silently, easily
740 2017-07-17 18:41:22	0|ryanofsky|sipa, maybe an example of where query-string would not work & uri-path would work?
741 2017-07-17 18:41:56	0|sipa|18:10:59 < sipa> i guess it isn't worse - as long as everything you'd put in the path becomes a query parameter and nothing else
742 2017-07-17 18:42:00	0|ryanofsky|by the way, one reason i am more inclined to named wallet param is that i think it is actually better for safety
743 2017-07-17 18:42:00	0|sipa|ryanofsky: there isn't
744 2017-07-17 18:42:30	0|jtimon|sipa: oh, ok, then that's where I was confused! what are we discussing then?
745 2017-07-17 18:42:40	0|sipa|jtimon: JSON RPC argument vs URI
746 2017-07-17 18:42:53	0|sipa|jtimon: i'm not talking about URI path vs URI query string
747 2017-07-17 18:43:08	0|ryanofsky|named arguments are safer than positional arguments, because harder to screw up ordering, explicitly specifying wallet is better than not because easy to mess up port numbers
748 2017-07-17 18:43:42	0|ryanofsky|so i tend to think even with multiprocess wallet, param has utility
749 2017-07-17 18:44:37	0|jtimon|sipa: right, so there's no difference between JSON RPC argument vs URI, right?
750 2017-07-17 18:44:47	0|sipa|jtimon: yes there is
751 2017-07-17 18:45:05	0|sipa|jtimon: with URI based approaches, we can avoid the need for applications to know what wallet they're talking to
752 2017-07-17 18:45:13	0|sipa|or even know that there is such a thing as multiwallet
753 2017-07-17 18:45:23	0|jtimon|I thought you just said there isn't an example of what ryanofsky was asking for?
754 2017-07-17 18:45:31	0|jtimon|I'm getting more confused, sorry
755 2017-07-17 18:45:38	0|sipa|jtimon: sigh... that was about URI path vs URI query
756 2017-07-17 18:45:43	0|jtimon|I'll read the full conversation
757 2017-07-17 18:46:10	0|ryanofsky|jtimon, talking about 3 different things. (1) wallet in jsonrpc param (2) wallet in uri query param (3) wallet in uri-path
758 2017-07-17 18:46:14	0|jtimon|I don't think ryanofsky was asking anything about URI query, I know I wasn't
759 2017-07-17 18:46:19	0|BlueMatt|sipa: but thats my point...we cant cause none of the client libraries support it, and they're all like barely maintained last I heard
760 2017-07-17 18:46:37	0|sipa|ryanofsky: every JSON-RPC library supports passing in a URI
761 2017-07-17 18:46:39	0|sipa|eh, BlueMatt ^
762 2017-07-17 18:46:52	0|sipa|BlueMatt: the bitcoin specific shims can be trivially patched to pass that through
763 2017-07-17 18:47:29	0|jtimon|ryanofsky: and who is defending 2 ?
764 2017-07-17 18:47:29	0|sipa|adding a named wallet arg everywhere is far harder, and requires logic that imho is totally unneeded at that level
765 2017-07-17 18:48:01	0|jtimon|ryanofsky: I really thought we were only duscussing 1 vs 3 already, sorry
766 2017-07-17 18:48:38	0|sipa|jtimon: the main discussion is (1) vs (3)... though ryanofsky has suggested that he'd prefer (2) over (3)
767 2017-07-17 18:48:41	0|ryanofsky|i'm defending 2. i think there are practical tradeoffs between 1 & 2, but that 3 has no practical advantages over either and has disadvange of introducing undocumented half-baked url scheme
768 2017-07-17 18:48:44	0|BlueMatt|sipa: you could add a shim that passes the wallet arg in everywhere, too, which is already supported by some rpc client libraries, and all of this debate is useless if we have a process split anyway, cause then its about port numbers anyway
769 2017-07-17 18:49:01	0|BlueMatt|at least then we wont have new endpoints to maintain, just an extra arg that we can ignore
770 2017-07-17 18:49:06	0|sipa|BlueMatt: port numbers are also part of the URI
771 2017-07-17 18:49:12	0|jtimon|ok, I missed that, as said I will read the beginning of the discussion...
772 2017-07-17 18:49:22	0|BlueMatt|yes, but are treated differently by some rpc client software :p
773 2017-07-17 18:49:27	0|sipa|?
774 2017-07-17 18:49:48	0|BlueMatt|well apparently at least python-bitcoinrpc does not support endpoints, but does, obviously, already support a different port to connect to
775 2017-07-17 18:49:54	0|BlueMatt|or so I'm told
776 2017-07-17 18:49:59	0|sipa|yes, but that's easy to fix
777 2017-07-17 18:50:15	0|sipa|and doesn't change that's better if client libraries don't need to know what wallet they're connected to
778 2017-07-17 18:50:35	0|jtimon|yeah, don't know python-bitcoinrpc but I can't imagine how it wouldn't be trivial to adapt either way
779 2017-07-17 18:50:48	0|ryanofsky|BlueMatt, python-bitcoinrpc doesn't support multiple endpoints currently, you have to choose one in advance if you want to write a test that uses multiple wallets for example. but jonas pr changes that
780 2017-07-17 18:51:08	0|BlueMatt|ohoh, wait, so it does support endpoints but you just need multiple AuthServiceProxy's for it?
781 2017-07-17 18:51:10	0|jtimon|afk
782 2017-07-17 18:51:12	0|BlueMatt|wait, that might change my view
783 2017-07-17 18:51:13	0|BlueMatt|ugh
784 2017-07-17 18:51:15	0|BlueMatt|i give up
785 2017-07-17 18:51:19	0|ryanofsky|BlueMatt, exactly
786 2017-07-17 18:51:31	0|BlueMatt|oh, well i mean thats kinda more analygous to the port number changes :(
787 2017-07-17 18:51:34	0|BlueMatt|lol, sorry sipa
788 2017-07-17 18:52:00	0|sipa|ryanofsky: i think that you do bring up a good point that there is a risk in creation a half-baked url scheme
789 2017-07-17 18:52:12	0|ryanofsky|sipa, that is why i am fine with 2
790 2017-07-17 18:52:34	0|sipa|ryanofsky: i believe (2) has more risk in going the wrong way than (3)
791 2017-07-17 18:53:03	0|sipa|and i guess part of that argument is aesthetics
792 2017-07-17 18:53:06	0|ryanofsky|i also want to point out that there we are talking about 1 practical tradeoff between 1 & 2/3, there are other practical advantages to 1 like not having to modify bitcoin-cli
793 2017-07-17 18:53:15	0|BlueMatt|can we use the lsb of the next blockhash?
794 2017-07-17 18:54:04	0|ryanofsky|sipa, ok that is what i think i'm not understanding. is there more to the argument than aesthetics...
795 2017-07-17 18:54:05	0|sipa|ryanofsky: i am fine with (1) over (2)/(3) if we believe the separation isn't thought out enough
796 2017-07-17 18:54:41	0|sipa|ryanofsky: it's a slippery slope argument, so you can counter anything i say with "well, we can just decide not to do that"
797 2017-07-17 18:55:07	0|sipa|ryanofsky: but i believe there is a risk that (2) will make it too easy for us to say "oh, let's put the currency unit in the URI"
798 2017-07-17 18:55:10	0|ryanofsky|sipa, not sure i understand an example of something bad....
799 2017-07-17 18:55:31	0|ryanofsky|you were saying some other parameters are bad to put in query strings?
800 2017-07-17 18:56:53	0|sipa|actually, i change my mind - that wouldn't be bad
801 2017-07-17 18:57:37	0|ryanofsky|also just to list other practical advantages i see in (1) over (2/3). Requires no changes to bitcoin-cli. Is properly documented, easy to understand, just a param. Encourages named arguments. Allows checking wallet name for safety even with multiprocess.
802 2017-07-17 18:58:09	0|sipa|so yes, my preference for (3) over (2) is aesthetics (and, _if_ we have a well thought-out separation, i also think (2) does not really have advantages over (3))
803 2017-07-17 18:59:22	0|ryanofsky|agree maybe (2) may not have advantages over (3) if you have a well-thought uri-path schema
804 2017-07-17 18:59:52	0|sipa|and things like a currency type or timeout can still be query string options rather than paths
805 2017-07-17 19:00:21	0|ryanofsky|i definitely have opposite aesthetics though, straight key=value named argument vastly preferable to me to inflexible /positional/path/stuff
806 2017-07-17 19:01:28	0|sipa|well, (purely aesthetics argument), paths feel like "i am now talking to a different subsystem!"; query options feel like "here is some extra stuff that may or may not affect how you're interpreting my request"
807 2017-07-17 19:02:55	0|ryanofsky|sure, i can see that
808 2017-07-17 19:03:30	0|sipa|and given an intention (if we agree to that) to move the wallet out, i think it makes sense to treat it as a different subsystem
809 2017-07-17 19:03:41	0|morcos|at the risk of derailing this convo on the brink of agreement, sipa: achow: do you still want #10579 for 0.15, it doesn't look as far along as 10571
810 2017-07-17 19:03:42	0|gribble|https://github.com/bitcoin/bitcoin/issues/10579 | [RPC] Split signrawtransaction into wallet and non-wallet RPC command by achow101 · Pull Request #10579 · bitcoin/bitcoin · GitHub
811 2017-07-17 19:03:42	0|ryanofsky|maybe i don't understand what you see in multiprocess world. i see wallet=filename still being something you can specify for safety
812 2017-07-17 19:04:17	0|ryanofsky|but you are seeing us do this elaborate /v1/wallet/ /v1/node thing and then the /v1/wallet stuff goes in the trash because it is no longer a "separate subsystem"?
813 2017-07-17 19:04:25	0|morcos|oops achow101 ^
814 2017-07-17 19:04:55	0|sipa|ryanofsky: no, /v1/wallet would remain - the wallet process just would not expose /v1/node anymore
815 2017-07-17 19:05:16	0|ryanofsky|sipa, ok. that is aesthetically ugly to me :)
816 2017-07-17 19:05:25	0|sipa|ryanofsky: please clarify
817 2017-07-17 19:06:39	0|sipa|(as in, i'd genuinely interested in hearing why)
818 2017-07-17 19:06:41	0|sipa|*i'm
819 2017-07-17 19:06:53	0|ryanofsky|wallet filename at that point is no longer "specifying a subsystem". it is just redundant at that point. we have to go on treating it as this magical thing different from other parameters forever
820 2017-07-17 19:07:13	0|sipa|ryanofsky: well i expect one process to remain capable of handling multiple wallets
821 2017-07-17 19:07:50	0|sipa|a lightweight node is far cheaper than a full node, but handling an individual wallet compared to that is still orders of magnitude less
822 2017-07-17 19:08:22	0|achow101|morcos: I would like to have it in 0.15 but it hasn't been getting any review
823 2017-07-17 19:09:19	0|sipa|ryanofsky: actually, no
824 2017-07-17 19:09:43	0|sipa|ryanofsky: the whole point of having it in a URI is that it shouldn't be treated as part of the interface
825 2017-07-17 19:10:32	0|sipa|someone could create a new lightweight implementation that has the same API (unlikely, i know), which only exposes what we have now under /v1/wallet/[walletname], but just exposes it as '/blah'
826 2017-07-17 19:12:40	0|jnewbery|sorry, I was away from my desk. Seems like I missed out on all the fun.
827 2017-07-17 19:12:42	0|sipa|however, if people feel that we haven't thought through the implication of that, and whether we can do that separation cleanly... we should just do (1)
828 2017-07-17 19:12:54	0|jnewbery|At the risk of going over old ground, I prefer (3) for a couple of reasons:
829 2017-07-17 19:13:21	0|jnewbery|1. each wallet is conceptually a separate resource, so it makes sense to me to have different URIs
830 2017-07-17 19:13:40	0|jnewbery|That's true whether or not we go for wallet separation in future
831 2017-07-17 19:14:21	0|jnewbery|*wallet process separation
832 2017-07-17 19:14:23	0|ryanofsky|thanks jnewbery, you are finally writing the documentatino i was asking for :)
833 2017-07-17 19:14:58	0|jnewbery|2. It offers a smoother upgrade path for clients if we do go for wallet process separation and each wallet binds to its own port
834 2017-07-17 19:15:27	0|jnewbery|namely: change the endpoint now to /wallet/<blah>, and then change the endpoint later to <wallet port>
835 2017-07-17 19:16:13	0|jnewbery|I'm *much* less worried than ryanofsky about implementing a uri scheme now that we want to get rid of later. This should all be considered unstable/experimental anyway
836 2017-07-17 19:16:36	0|jnewbery|yes, I'll happily write documentation
837 2017-07-17 19:17:20	0|sipa|ryanofsky: if anything - thanks for discussing this; it's made your point clearer to me, and helped me understand my own preferences better
838 2017-07-17 19:17:49	0|ryanofsky|your second reason is reason sipa & gmaxwell like approaches (2)&(3) over approach (1), in that respect there is no distiniction between (2)/(3), and countervailing tradeoffs like having to modify bitcoin-cli and other things i mentioned above
839 2017-07-17 19:18:39	0|ryanofsky|and yes, it is clear that i am wayyy more concerned with backwards compatibility than other people
840 2017-07-17 19:19:16	0|jnewbery|In general, I'm also concerned about backwards compatibility. But in this specific case, I'm not
841 2017-07-17 19:19:21	0|ryanofsky|i'm probably just an outlier in that respect
842 2017-07-17 19:20:08	0|instagibbs|jnewbery, a brief(!) recap in the PR itself on the design choice being made would be nice for review/historicity sake
843 2017-07-17 19:21:08	0|jnewbery|> countervailing tradeoffs like having to modify bitcoin-cli
844 2017-07-17 19:21:25	0|ryanofsky|yes named arguments require no changes to bitcoin-cli
845 2017-07-17 19:21:39	0|sipa|ok, i just looked over #10849, and it is suffuciently simple that i don't think i care about implementation complexity compared to 10829
846 2017-07-17 19:21:40	0|gribble|https://github.com/bitcoin/bitcoin/issues/10849 | Multiwallet: simplest endpoint support by jonasschnelli · Pull Request #10849 · bitcoin/bitcoin · GitHub
847 2017-07-17 19:22:16	0|jnewbery|I think (2) is better in that respect. We need to modify bitcoin-cli, but we have those changes already coded. With (1), every script that calls wallet methods using bitcoin-cli needs to be changed because positional arguments are no longer supported
848 2017-07-17 19:22:25	0|jnewbery|or have I misunderstood?
849 2017-07-17 19:22:53	0|sipa|jnewbery: that is also a good argument
850 2017-07-17 19:23:22	0|ryanofsky|jnewbery, you understood. i see that is an argument the other way because named arguments are safer
851 2017-07-17 19:24:23	0|jnewbery|with (2), the bitcoin-cli caller needs to change to include a `-usewallet` argument. With (1) the entire invocation needs to change
852 2017-07-17 19:24:37	0|ryanofsky|jnewbery, correct
853 2017-07-17 19:25:00	0|jnewbery|so with (2), it's just as safe. There's no default wallet - it has to be explicitly specified
854 2017-07-17 19:25:19	0|ryanofsky|i'm talking about general safety of named parameters vs positional arguments
855 2017-07-17 19:25:52	0|ryanofsky|i don't think discouraging positional arguments is bad, i actually think it is good
856 2017-07-17 19:25:59	0|jnewbery|oh, absolutely agree there - everyone should use named. But that's an orthogonal point
857 2017-07-17 19:26:03	0|sipa|ryanofsky: that seems orthogonal
858 2017-07-17 19:26:06	0|ryanofsky|i think changes to bitcoin-cli are simple but ugly
859 2017-07-17 19:26:11	0|ryanofsky|anyway these are minor points
860 2017-07-17 19:26:31	0|ryanofsky|i think we all understand tradeoffs between 1 / 2&3 at this point?
861 2017-07-17 19:27:44	0|ryanofsky|for those who have clear notions of what "conceptually a separate resource" means & who don't care about backwards compatibility, there is no difference between 2&3
862 2017-07-17 19:29:12	0|sipa|for those who don't have clear notions, you mean?
863 2017-07-17 19:29:57	0|ryanofsky|i'm saying i don't know clearly what "separate resource" means. it just seems like an arbitrary distinction
864 2017-07-17 19:30:56	0|ryanofsky|it just seems weird to me that you'd want to structure a path scheme around wallet filename, but if you're confident that this is the way to go, then great
865 2017-07-17 19:31:04	0|sipa|how about this: all parts of a request that are expected to be identical between all calls made by a single client application should go into a path
866 2017-07-17 19:31:46	0|jnewbery|Yes, I think we all understand the tradeoffs. My preference order is still 3 > 2 > 1, but I'm happy with any of them going in now, and then coming up with a stable design for all of this in time for v0.16.
867 2017-07-17 19:32:30	0|ryanofsky|sipa, to me it's ugly you even want to make that distinction. aesthetically i prefer if all parts of request should just be treated as similarly as possible
868 2017-07-17 19:33:27	0|sipa|ryanofsky: actually, that sentence applies to (2) and (3) equally
869 2017-07-17 19:33:43	0|sipa|so replace 'path' with 'uri' in it
870 2017-07-17 19:33:46	0|ryanofsky|there are practical reasons (imo weak ones for wanting to make wallet special enough to go in url rather than json request). but making it root of brand new uri schema seems overboard to me
871 2017-07-17 19:34:01	0|jtimon|sipa: I guess we don't move "jsonrpc": "1.0" to the path because it's part of the rpc scheme specification or something?
872 2017-07-17 19:34:27	0|sipa|jtimon: no, it's part of the JSON-RPC spec
873 2017-07-17 19:34:38	0|sipa|oh, that's what you mean; yes
874 2017-07-17 19:34:42	0|ryanofsky|in (1) wallet is just one of many normal params. in (2) wallet is special enough to be a url param. in (3) wallet is root of a new uri-scheme
875 2017-07-17 19:34:51	0|jtimon|right, that's what I meant, but didn't rename json-rpc name
876 2017-07-17 19:35:19	0|ryanofsky|uri-path scheme i mean
877 2017-07-17 19:35:22	0|jtimon|wouldn't putting the wallet there violate the json-rpc spec ?
878 2017-07-17 19:35:30	0|sipa|no
879 2017-07-17 19:35:40	0|sipa|it's just a different URL you're using
880 2017-07-17 19:37:56	0|jtimon|no, I mean, putting wallet in the json data alongside method, jsonrpc, params and id
881 2017-07-17 19:38:15	0|ryanofsky|jtimon, that isn't (1) (2) or (3).
882 2017-07-17 19:38:16	0|sipa|that's not what's being suggested
883 2017-07-17 19:38:26	0|ryanofsky|(1) is just sticking it into params
884 2017-07-17 19:38:29	0|sipa|(1) suggests making the wallet part of 'params'
885 2017-07-17 19:38:33	0|jtimon|oh, I see
886 2017-07-17 19:57:26	0|promag|BlueMatt: saw some rpc functions, all have spaces in ()
887 2017-07-17 19:58:46	0|BlueMatt|promag: hmm, ok
888 2017-07-17 20:00:18	0|morcos|wumpus: not sure where we are with feature freeze, but i think #10707 is going to make it, looks ready for merge
889 2017-07-17 20:00:20	0|gribble|https://github.com/bitcoin/bitcoin/issues/10707 | Better API for estimatesmartfee RPC by morcos · Pull Request #10707 · bitcoin/bitcoin · GitHub
890 2017-07-17 20:00:29	0|morcos|#10672 can also be merged
891 2017-07-17 20:00:31	0|gribble|https://github.com/bitcoin/bitcoin/issues/10672 | Avoid division by zero in the case of a corrupt estimates file by practicalswift · Pull Request #10672 · bitcoin/bitcoin · GitHub
892 2017-07-17 20:02:14	0|BlueMatt|we should probably pull an 0.14 and just say that we're frozen with an exception for things already tagged that make it in the next day or two?
893 2017-07-17 20:03:53	0|morcos|someone PLEASE tag #10830
894 2017-07-17 20:03:54	0|gribble|https://github.com/bitcoin/bitcoin/issues/10830 | [WIP] [wallet] keypool restore by jnewbery · Pull Request #10830 · bitcoin/bitcoin · GitHub
895 2017-07-17 20:04:00	0|morcos|I keep forgetting about it b/c its not on the milestone list
896 2017-07-17 20:04:26	0|morcos|sipa: aren't you the one that was threatening to rip the entire wallet out if we didn't get a version of that in
897 2017-07-17 20:04:37	0|sipa|morcos: lol
898 2017-07-17 20:04:58	0|sipa|tagged
899 2017-07-17 20:05:03	0|morcos|thanks!
900 2017-07-17 20:05:08	0|jnewbery|I'm working on a cut-down version of #10830
901 2017-07-17 20:05:10	0|gribble|https://github.com/bitcoin/bitcoin/issues/10830 | [WIP] [wallet] keypool restore by jnewbery · Pull Request #10830 · bitcoin/bitcoin · GitHub
902 2017-07-17 20:05:20	0|jnewbery|probably under a different PR number - sorry!
903 2017-07-17 20:05:43	0|jnewbery|I think the prevailing view is we don't need the node sync pause stuff for 0.15
904 2017-07-17 20:06:16	0|sipa|jnewbery: you'd need to do something, though
905 2017-07-17 20:06:20	0|sipa|what is your suggestion?
906 2017-07-17 20:07:10	0|jnewbery|I'm trying to understand what's required. You've mentioned before that HD split makes things worse, but I can't understand why that would be true
907 2017-07-17 20:07:35	0|sipa|it doesn't; i was confused
908 2017-07-17 20:07:47	0|jtimon|so, regarding #9806 txoutsbyaddress, txoutsbyscript and all that...what are the plans for 0.15 again (if any), it seems I may have misunderstood things there
909 2017-07-17 20:07:48	0|jnewbery|ok, good
910 2017-07-17 20:07:49	0|gribble|https://github.com/bitcoin/bitcoin/issues/9806 | txoutsbyaddress index (take 3) by droark · Pull Request #9806 · bitcoin/bitcoin · GitHub
911 2017-07-17 20:08:15	0|jnewbery|so, I'll keep the stuff that marks all keys up to a used key as used
912 2017-07-17 20:08:56	0|jnewbery|gmaxwell says "we could couple that with something that prolongs keeping an encrypted wallet unlocked while syncing/rescanning is running". I'm just looking at how locking/unlocking works to understand what's required there
913 2017-07-17 20:09:46	0|sipa|i think you can just shutdown when the keypool goes below some threshold
914 2017-07-17 20:10:11	0|sipa|which can only happen when you're recovering from an old wallet backup anyway
915 2017-07-17 20:10:51	0|jnewbery|hows that? What if your wallet is encrypted and you can't top up?
916 2017-07-17 20:11:12	0|sipa|then you have a problem
917 2017-07-17 20:11:33	0|sipa|because you're going to silently miss transactions
918 2017-07-17 20:12:54	0|jnewbery|right. Sorry - I don't understand how your keypool can only go below a threshold if you're recovering from an old wallet backup
919 2017-07-17 20:13:32	0|sipa|during normal operation, you never see a key used on the network that you didn't create with getnewaddress, which will make you top up
920 2017-07-17 20:16:24	0|jnewbery|ok, so I get a bunch of adddresses with getnewaddress, I hand them out, my wallet locks, and then I start seeing transactions to those addresses. What next? Isn't my unused keypool running down as I see those transactions?
921 2017-07-17 20:16:45	0|sipa|jnewbery: getnewaddress already marked those keys as used
922 2017-07-17 20:17:00	0|sipa|it can refuse to give you a new one before topping up
923 2017-07-17 20:17:05	0|jnewbery|ah, ok
924 2017-07-17 20:17:13	0|sipa|it does now, but the threshold is 0
925 2017-07-17 20:17:19	0|sipa|that's probably too low, but easy to change
926 2017-07-17 20:17:51	0|sipa|if you're able to avoid hitting a keypool of 0 when its maximum is 100, you're certainly able to avoid hitting 100 when the default is 1000
927 2017-07-17 20:18:04	0|jnewbery|ok, I'll take a look. Thanks
928 2017-07-17 20:21:41	0|promag|https://github.com/bitcoin/bitcoin/pull/9502 it's supposed to keep syncing headers? sipa?
929 2017-07-17 20:22:23	0|sipa|promag: i'm not sure what it's supposed to do, but i believe that is what it does yes
930 2017-07-17 20:30:59	0|BlueMatt|sipa: can you update the state of #10526?
931 2017-07-17 20:31:00	0|gribble|https://github.com/bitcoin/bitcoin/issues/10526 | Force on-the-fly compaction during pertxout upgrade by sipa · Pull Request #10526 · bitcoin/bitcoin · GitHub
932 2017-07-17 20:31:35	0|BlueMatt|can we confirm if we need it for 15 or not?
933 2017-07-17 20:34:58	0|sipa|BlueMatt: willdo
934 2017-07-17 20:38:53	0|jnewbery|sipa: if we shutdown when the keypool goes below a threshold, how would I start with an old encrypted wallet that has fewer than the minimum threshold keypool? It'll shutdown before I have a change to unlock the wallet
935 2017-07-17 20:40:33	0|sipa|jnewbery: i don't know
936 2017-07-17 20:54:16	0|jtimon|btw, with gettxout I think it should either include confirmed utxos when calling with include_mempool or have include_mempool renamed to mempool_only or something of the sort. At the very very least s/Whether to include the mempool/Only search in the mempool. Default: true/
937 2017-07-17 20:54:19	0|jtimon|thoughts ?
938 2017-07-17 20:55:13	0|sipa|jtimon: include_mempool is the wrong name
939 2017-07-17 20:55:32	0|sipa|it either presents the view of the utxo at the last block
940 2017-07-17 20:55:44	0|sipa|or the one as seen by the mempool
941 2017-07-17 20:56:50	0|jtimon|right, my point is that if we want to maintain the name, we can first consider the mempool, if not, the current utxo; and the caller can check whether "confirmations" == 0 or not
942 2017-07-17 20:57:13	0|jtimon|but I take it as you prefer just renaming, right?
943 2017-07-17 20:57:23	0|sipa|or properly documenting
944 2017-07-17 20:57:44	0|sipa|but the RPC is about viewing the utxo set... not random access to txouts
945 2017-07-17 20:58:12	0|sipa|we have two utxo sets... the one defined by the blockchain, and the one defined by the blockchain+mempool
946 2017-07-17 21:00:17	0|jnewbery|sipa: I can think of a couple of good solutions to the old encrypted wallet at start: 1. have a bitcoin-wallet util that can topup the keypool offline. 2. have a `loadwallet` RPC that can decrypt on load
947 2017-07-17 21:00:47	0|jnewbery|obviously neither of those are in v0.15. It seems a shame to merge something that could make an encrypted wallet file unloadable
948 2017-07-17 21:02:00	0|jnewbery|there would be one way to recover a blocked wallet: invalidateblock at the wallet's best block, load and unlock the wallet, topup keypool, then reconsiderblock. But I don't think we should be telling users to do that!
949 2017-07-17 21:02:42	0|jtimon|sipa: mhmm, but a confirmed utxo won't appear if you call with include_mempool
950 2017-07-17 21:03:25	0|sipa|jtimon: of course not
951 2017-07-17 21:03:38	0|sipa|it is not an unspent output, when looking at the mempool
952 2017-07-17 21:03:49	0|sipa|wait
953 2017-07-17 21:03:55	0|sipa|i'm not sure what you're saying
954 2017-07-17 21:04:55	0|sipa|jtimon: my answer is about a transaction that is unspent in the chain, but spent by a mempool tx
955 2017-07-17 21:05:16	0|jtimon|right, that won't appear if you chose include_mempoo=true
956 2017-07-17 21:05:45	0|sipa|yes, intentionally
957 2017-07-17 21:05:50	0|jtimon|ok
958 2017-07-17 21:06:03	0|sipa|because it is not an unspent output when looking at the mempool
959 2017-07-17 21:07:10	0|jtimon|oh, I see, but if it's both spent in the chain and the mempool it will appear, I was missing that, thanks
960 2017-07-17 21:07:40	0|sipa|unspent, you mean
961 2017-07-17 21:07:59	0|jtimon|perhaps we should test that too
962 2017-07-17 21:08:07	0|jtimon|yeah, unspent sorry
963 2017-07-17 21:08:40	0|jtimon|ok, so I think I'll write a PR correcting the documentation and testing that case
964 2017-07-17 21:09:10	0|sipa|cool
965 2017-07-17 21:10:36	0|jtimon|but before closing #10822, I would like to know more about what the general thoughts about gettxoutbyaddress, gettxoutbyscript and all that
966 2017-07-17 21:10:37	0|gribble|https://github.com/bitcoin/bitcoin/issues/10822 | RPC: Also serve txo from gettxout (not just utxo and mempool) by jtimon · Pull Request #10822 · bitcoin/bitcoin · GitHub
967 2017-07-17 21:11:21	0|sipa|jtimon: i believe other people have different opinions, but imho that does not belong in core
968 2017-07-17 21:12:08	0|sipa|or at least only if we modularize the code enough so that it's a totally separately pluggable thing
969 2017-07-17 21:14:27	0|jtimon|by thought was to have something like -scriptpubkeyindex analogous to txindex or something like that, but I see you don't like having -txindex already
970 2017-07-17 21:14:47	0|sipa|yes
971 2017-07-17 21:15:12	0|sipa|i'm very strongly opposed to any functionality that requires having a fully indexed blockchain
972 2017-07-17 21:15:17	0|jtimon|of course I don't need this to be in core for my purposes, but since people were talking about it I was trying to find synergies
973 2017-07-17 21:15:34	0|sipa|if it's just the utxo set... it's less bad, but i still think it's beyond our scope
974 2017-07-17 21:16:01	0|jtimon|but a -scriptpubkeyindex only for current utxo would be more acceptable?
975 2017-07-17 21:16:07	0|jtimon|ok
976 2017-07-17 21:18:16	0|jtimon|thanks, I will ask on the next meeting about if I haven't closed #10822 by the time if more people give me their opinion
977 2017-07-17 21:18:17	0|gribble|https://github.com/bitcoin/bitcoin/issues/10822 | RPC: Also serve txo from gettxout (not just utxo and mempool) by jtimon · Pull Request #10822 · bitcoin/bitcoin · GitHub
978 2017-07-17 21:19:12	0|sipa|i think we're a bit too busy with 0.15 :)
979 2017-07-17 21:38:54	0|sipa|jnewbery: yeah, i don't know
980 2017-07-17 21:39:12	0|sipa|a cleaner solution is to actually support stopping a sync, and allowing it to continue with a keypool unlock
981 2017-07-17 21:39:40	0|sipa|but this is an unusual situation, which will only be reached when recovering from a backup anyway
982 2017-07-17 21:41:21	0|jcorgan|just fyi, we (gnuradio project) just got a github PR with a malware PDF attached to a comment, looked automated
983 2017-07-17 21:41:55	0|jcorgan|well, assuming it is malware, virustotal found 0 hits
984 2017-07-17 21:44:34	0|sipa|jcorgan: good to know, thanks
985 2017-07-17 21:51:47	0|jtimon|sipa: yeah, sure, no hurry on my part, I just thought something re gettxoutbyaddress for 0.15, I guess it was just for after 0.15
986 2017-07-17 22:13:04	0|bitcoin-git|13bitcoin/06master 148276e70 15Chris Stewart: Adding assert to avoid a memory access violation inside of PartialMerkleTree::CalcHash()...
987 2017-07-17 22:13:04	0|bitcoin-git|[13bitcoin] 15sipa pushed 2 new commits to 06master: 02https://github.com/bitcoin/bitcoin/compare/0b019357ff09...fee0d803fb55
988 2017-07-17 22:13:05	0|bitcoin-git|13bitcoin/06master 14fee0d80 15Pieter Wuille: Merge #9980: Fix mem access violation merkleblock...
989 2017-07-17 22:13:18	0|bitcoin-git|[13bitcoin] 15sipa closed pull request #9980: Fix mem access violation merkleblock (06master...06fix_mem_access_violation_merkleblock) 02https://github.com/bitcoin/bitcoin/pull/9980
990 2017-07-17 22:35:50	0|bitcoin-git|[13bitcoin] 15achow101 opened pull request #10857: [RPC] Add a deprecation warning to getinfo's output (06master...06deprecate-getinfo) 02https://github.com/bitcoin/bitcoin/pull/10857
991 2017-07-17 23:14:52	0|bitcoin-git|[13bitcoin] 15achow101 opened pull request #10858: [RPC] Add "errors" field to getblockchaininfo and unify "errors" field in get*info RPCs (06master...06getblockchaininfo-errors) 02https://github.com/bitcoin/bitcoin/pull/10858
992 2017-07-17 23:54:51	0|bitcoin-git|[13bitcoin] 15sipa pushed 3 new commits to 06master: 02https://github.com/bitcoin/bitcoin/compare/fee0d803fb55...75b5643c47c3
993 2017-07-17 23:54:52	0|bitcoin-git|13bitcoin/06master 1406bcdb8 15Alex Morcos: Convert named argument from nblocks to conf_target...
994 2017-07-17 23:54:52	0|bitcoin-git|13bitcoin/06master 14439c4e8 15Alex Morcos: Improve api to estimatesmartfee...
995 2017-07-17 23:54:53	0|bitcoin-git|13bitcoin/06master 1475b5643 15Pieter Wuille: Merge #10707: Better API for estimatesmartfee RPC...
996 2017-07-17 23:55:16	0|bitcoin-git|[13bitcoin] 15sipa closed pull request #10707: Better API for estimatesmartfee RPC  (06master...06bettersmartfeeapi) 02https://github.com/bitcoin/bitcoin/pull/10707