19:00:19 #startmeeting 19:00:19 Meeting started Thu Oct 11 19:00:19 2018 UTC. The chair is wumpus. Information about MeetBot at http://wiki.debian.org/MeetBot. 19:00:19 Useful Commands: #action #agreed #help #info #idea #link #topic. 19:00:23 #bitcoin-core-dev Meeting: wumpus sipa gmaxwell jonasschnelli morcos luke-jr btcdrak sdaftuar jtimon cfields petertodd kanzure bluematt instagibbs phantomcircuit codeshark michagogo marcofalke paveljanik NicolasDorier jl2012 achow101 meshcollider jnewbery maaku fanquake promag provoostenator 19:00:27 hi 19:00:28 hi 19:00:33 hello! 19:01:24 proposed topics? 19:01:30 Hi. 19:02:11 hp? 19:02:31 topic: OSX dmg issue 19:02:37 +proposal 19:02:49 * jonasschnelli is in train, connection may randomly drop 19:03:15 IRCing from a train is pretty hard-core 19:03:49 #topic High priority for review 19:03:50 well... 19:04:08 https://github.com/bitcoin/bitcoin/projects/8 19:05:08 jonasschnelli: it's better than it used to be, I suppose :D 19:05:42 (they have 4g repeater in almost all trains now, probably expect the one im sitting in) 19:07:12 I used to travel by train a lot ~10 years ago, with a crappy GPRS modem, tunneling over ssh, losing connection every few minutes, it was really a nightmare at the time; was before trains had WiFi here in NL 19:07:33 any suggestions to add to high priority for review? 19:07:35 #14291 could some review by wallet "owners" 19:07:37 https://github.com/bitcoin/bitcoin/issues/14291 | wallet: Add ListWalletDir utility function by promag · Pull Request #14291 · bitcoin/bitcoin · GitHub 19:08:01 you already have 14291 on the list 19:08:22 I know 19:08:26 `hi 19:09:14 hi 19:09:46 #topic OSX dmg issue (jonasschnelli) 19:09:48 hi 19:10:00 Issue is #14414 19:10:01 https://github.com/bitcoin/bitcoin/issues/14414 | Problem upgrading to 0.17.0 on macOS High Sierra 10.13.6 · Issue #14414 · bitcoin/bitcoin · GitHub 19:10:13 TLDR: our 0.17.0 DMG image is corrupt... 19:10:30 Probably due to a upstream "bug" in the ds_store python module 19:11:07 IMO the best way to fix this is by "rolling back" to manual generated .DS_Store files 19:11:09 it *crashes* the finder? 19:11:27 wumpus: in one case, I had a finder crash. Yes. 19:11:35 But most cases, it just didn't open the image 19:11:36 me too 19:11:44 The crash worries me a bit... 19:11:45 that's pretty interestin if it's exploitable 19:11:49 only one crash though 19:12:09 anyhow, how did we get here, why didn't it happen before? 19:12:26 The DS_Store fileformat is not open.. 19:12:35 it has been reverse engeneered AFAIK 19:12:42 can we go back to an older ds_store python module 19:12:48 and the module we use (ds_store module in python) probably messes up 19:12:58 yes 19:13:00 and the mess up only appears on OSX 10.12 upwards 19:13:11 I don't think older versions would work.. 19:13:26 I don't think we have upgraded from an older version 19:13:30 For me it crashes the Finder always. Like 100% reproducibility. 19:13:32 but it used to work right? 19:13:42 only on older OSX versions 19:13:53 so 0.16.x also crashes on newer OSX? 19:13:56 I think its an "incompatibility" that only affects OSX 10.12+ 19:14:17 wumpus: good question. 19:14:28 I mean the issue talks about upgrading to 0.17 19:14:32 I think they don't... which would lead to something we changed. 19:15:12 I don't think much changed with regard to macosx tooling since then but I guess it's worth checking 19:15:35 Maybe the move to Bionic? 19:15:49 do we use it from the distribution? 19:16:03 no... we load the python package 19:16:13 but maybe its not the ds_store. packaghe 19:16:14 can we know the module version used in a previous working version? 19:16:27 AFAIK we have not touched it,... 19:16:31 it's possible that something else in the dependency forest causes a subtle change, of course 19:16:46 Just tried with 0.16.3 and it doesn't crash here. So it's a regression for sure. 19:16:48 maybe a newer python 19:16:51 0.16.3 is not affected (tested just now) 19:17:22 Removing the DS_Store file will cure the issue though... 19:17:29 could be the DMG creation 19:17:37 can just *remove* it? 19:17:56 wumpus: I propose to just replace it with a manually created DS_Store... 19:17:58 huh, both 0.17 and 0.16.3 use the same stuff right? 19:18:10 I... don't understand enough about OSX to say anything sensible about this 19:18:29 hi. 19:18:54 promag: 0.16.3 builds using trusty, 0.17 started building using bionic 19:19:15 However, .. I'll investigate further (didn't hat too much time last days) 19:19:20 might want to compare the ds_store files between both 19:19:21 python3? 19:19:31 it might be some unicode encoding issue or whatnot 19:19:44 The DS_Store file is just a file that contains info about the disk image background and icon position, etc. 19:19:45 0.16 was already python 3 wasn't it? 19:20:02 wumpus: good point.. i'll do that 19:20:20 But #11881 is not in 0.16 19:20:23 https://github.com/bitcoin/bitcoin/issues/11881 | Remove Python2 support by jnewbery · Pull Request #11881 · bitcoin/bitcoin · GitHub 19:20:39 jonasschnelli: if we can just remove it, that might be the best way forward, instead of relying on someone's reverse engineering which is apparently failing for new versions 19:20:55 anyway shame on finder.. 19:20:56 It changes custom_dsstore.py and macdeployqtplus 19:21:02 Yes. But the nice background is cool 19:21:07 (Its somehow expected on OSX) 19:21:36 ken2812221: is 11881 not 0.16 but 0.17? 19:21:41 I agree nice things are nice, but working at all is more important :) 19:21:53 sure. :) 19:22:22 Removing it is an option,.. lets wait a couple of more days to track the issue and if we can't fix it, remote it 19:22:42 (and wait until someome brings it back in a stable way) 19:22:57 we could generate it with trusty thru docker 19:22:58 11881 is indeed in 0.17 but not 0.16 19:23:47 (if that's the problem) 19:24:09 then it's probably 11881 19:25:10 we can try reverting it, at least for 0.17.x branch 19:25:31 but in any case we need to knwo what is the difference in the generated file 19:25:53 might be a straightforward difference or a nightmare 19:25:57 I'll test a revert in my WIP PR via gitian 19:28:17 thanks! hope a revert is doable by now, but I guess, there's not much change in those files 19:28:21 any other topics? 19:30:39 apparently not ! 19:30:54 thanks everyone, wrapping up the meeting then 19:31:03 #endmeeting