--- Log opened Thu Jul 18 00:00:16 2019 01:52 -!- jonatack [6de8e38a@109.232.227.138] has joined #bitcoin-builds 01:55 < jonatack> dongcarl: yay! \o/ 03:06 -!- jonatack [6de8e38a@109.232.227.138] has quit [Remote host closed the connection] 03:07 -!- jonatack [6de8e38a@109.232.227.138] has joined #bitcoin-builds 03:31 -!- jb55 [~jb55@gateway/tor-sasl/jb55] has quit [Ping timeout: 260 seconds] 03:37 -!- jonatack [6de8e38a@109.232.227.138] has quit [Ping timeout: 260 seconds] 07:11 -!- jonatack [6dca6b0f@109.202.107.15] has joined #bitcoin-builds 07:31 -!- jb55 [~jb55@gateway/tor-sasl/jb55] has joined #bitcoin-builds 07:37 -!- jonatack [6dca6b0f@109.202.107.15] has quit [Ping timeout: 260 seconds] 07:54 -!- hebasto [~hebasto@95.164.65.194] has joined #bitcoin-builds 09:42 -!- jonatack [58aba822@jau64-1-88-171-168-34.fbx.proxad.net] has joined #bitcoin-builds 09:57 -!- jonatack [58aba822@jau64-1-88-171-168-34.fbx.proxad.net] has quit [Ping timeout: 260 seconds] 11:13 < cfields> dongcarl: that all sounds great! 11:13 < cfields> Having a look now. 11:15 < cfields> Wait. This is just an arch package? 11:17 < cfields> We could cut/glue our own package together too; that's just cheating :) 11:17 < cfields> Or is there something upstream I'm not seeing? 11:18 < dongcarl> cfields: oh the xorgproto info is a little irrelevant now 11:18 < dongcarl> I changed tactics 11:18 < dongcarl> See the X pruning PR on the repo 11:18 < dongcarl> Sorry 😐 11:19 < cfields> Heh, ok. 11:36 < dongcarl> cfields: the X pruning PR is at the edge of my comfort zone, so would love your feedback there 11:37 < cfields> dongcarl: should I start there? 11:38 < dongcarl> cfields: yes that one would be the best! :-) 11:38 < dongcarl> https://github.com/bitcoin/bitcoin/pull/16408 11:40 < cfields> roger 11:44 < dongcarl> :-) 11:47 -!- jonatack [8613b3b3@134.19.179.179] has joined #bitcoin-builds 12:20 < dongcarl> cfields: Looked at your comments... 12:21 < dongcarl> Wondering how you'd handle: https://github.com/qt/qtbase/blob/81b29a44d248b5037a2f05dd8815e000d9472f0e/src/plugins/platforms/xcb/qxcbcursor.cpp#L52 12:21 < dongcarl> Would you just replace every `XC_` usage with the actual value? See: https://github.com/qt/qtbase/blob/81b29a44d248b5037a2f05dd8815e000d9472f0e/src/plugins/platforms/xcb/qxcbcursor.cpp#L392 12:22 < dongcarl> Here's cursorfont.h: https://github.com/freedesktop/xorg-libX11/blob/master/include/X11/cursorfont.h 12:26 < cfields> dongcarl: looks like all of cursorIdForShape is only needed for the libX11 fallback condition. 12:26 < cfields> so ifdef that function, then ifdef all of the "if (!cursor && cursorId)" blocks. 12:27 * dongcarl thinking 12:28 < dongcarl> Okay so right now `QT_CONFIG(xcb_xlib)` is false... 12:28 < cfields> (and then the include, obviously) 12:28 < dongcarl> so we'll just get the cursor from ` cursor = createNonStandardCursor(cshape);`? 12:30 < dongcarl> Okay yeah I think that makes sense... 12:30 < dongcarl> I love the rabbit holes that our build system leads me down 12:31 < cfields> dongcarl: I assume XCreateFontCursor is a libx11 function? 12:31 < dongcarl> checking... 12:32 < dongcarl> cfields: yup, from Xlib.h 12:33 < cfields> dongcarl: so either it's inline, or it's an unresolved export as I feared in the PR. 12:34 < dongcarl> cfields: Perhaps relevant, if `QT_CONFIG(xcb_xlib)` is true `QT_CONFIG(xlib)` has to be trrue 12:34 < cfields> So, I would much prefer to see patches instead :) 12:34 < cfields> At least that forces us to understand. 12:34 < dongcarl> I think given that fact... it's correct? 12:35 < cfields> Blah. Now I've forgotten how all that works again :p 12:35 < dongcarl> As in... during configure, qt checks for xlib first, and only checks for xcb_xlib if the xlib check succeeds 12:36 < cfields> In that case, none of this could should be being built...? 12:37 < cfields> *code 12:37 < dongcarl> I mean the `#if` block shouldn't be built... and L592-598 shouldn't be built... 12:37 < dongcarl> (in src/plugins/platforms/xcb/qxcbcursor.cpp) 12:37 < dongcarl> what else shouldn't be built? 12:38 < cfields> I thought we were chasing your compile errors. 12:39 < dongcarl> cfields: Oh, I'm just looking at the code and seeing how complex an upstream patch would be as you suggested :-) 12:39 < cfields> dongcarl: Sorry, let's back up... 12:39 * dongcarl backing up 12:40 < dongcarl> Just so we're on the same page, my PR as it is right now builds correctly, and `bitcoin-qt` runs properly 12:40 < cfields> dongcarl: add a dummy libx11 package that just dumps empty files where the necessary headers should be. 12:40 < dongcarl> cfields: Ohhhhhhh 12:40 < cfields> So that the "#include"s work, but you'll get compiler errors for whatever's actually missing. 12:40 * fanquake lurking 12:41 < dongcarl> cfields: And then fix the compile errors that i get for QT where they didn't properly `#if`? 12:41 < cfields> dongcarl: right. If we're lucky, only guards for the includes are missing. 12:42 < cfields> dongcarl: sorry, I thought you were trying to resolve those issues. Don't worry about code that's already properly guarded :) 12:42 < dongcarl> Cool, I'll play around with that and report back 12:46 < ryanofsky> MarcoFalke, appveyor broken? https://ci.appveyor.com/project/DrahtBot/bitcoin/builds/26082465 12:47 < ryanofsky> CMake Error at scripts/cmake/vcpkg_download_distfile.cmake:156 (message): Failed to download file. 12:48 < MarcoFalke> "EDITOR=true git commit --ammend" and then force push should rerun the build 12:48 < MarcoFalke> Oh, is is happening on all builds? 12:48 < ryanofsky> not sure, i saw it on two builds 12:49 < MarcoFalke> I'll reset the builds in a minute ... 12:49 < dongcarl> `EDITOR=true` 12:49 < MarcoFalke> Might be fixed by now 13:36 < MarcoFalke> Restarted by closing the pull request :/ 14:04 * dongcarl thinks he's eliminated `libX11` dependency with one patch but is not sure yet 14:26 -!- hebasto [~hebasto@95.164.65.194] has quit [Remote host closed the connection] 14:46 < dongcarl> cfields: Pushed 2 commits to that PR, eliminates `libX11`. 14:48 < cfields> dongcarl: hah! 14:48 < cfields> That's it? 14:48 < dongcarl> cfields: Yeah! I was surprised too... 14:48 < dongcarl> I did a full Guix build and ran bitcoin-qt... all the screens worked... 14:49 < cfields> That's great :) 14:49 < cfields> Need to look at the guards in detail, but yeah, if that's all it takes, it's a no-brainer. 14:50 < dongcarl> cfields: Yup, sounds good! Let me know if you find anything off about them 14:50 < cfields> dongcarl: well, for starters, are you sure they're all necessary? 14:51 < cfields> I thought it looked like some of that should've been guarded by xcb-xlib already. 14:54 < dongcarl> which one are you talking about exactly? 14:55 < cfields> dongcarl: heh, already popped back off my stack. I'll have to go refresh my memory. 14:55 < cfields> dongcarl: but, did you try _only_ adding the guard around the #include ? 14:56 < dongcarl> cfields: Hmmm no... Lemme try that... but I don't think it'll work because of `cursorIdForShape` 14:56 < cfields> you're probably right, I was thinking it was part of a larger guard. 14:58 < dongcarl> I think `cursorIdForShape` is part of the `#ifndef QT_NO_CURSOR` larger guard, but not any `QT_CONFIG(xcb_xlib) && QT_CONFIG(library)` guards 14:59 < dongcarl> Although I have no idea what `QT_CONFIG(library)` is... and just put it there to be safe... 15:03 < dongcarl> Yeah compiling shows it's definitely needed 15:03 < dongcarl> Perhaps I'll upstream and they'll give me feedback there 15:05 < cfields> https://github.com/qt/qtbase/blob/81b29a44d248b5037a2f05dd8815e000d9472f0e/src/plugins/platforms/xcb/qxcbcursor.cpp#L63 15:05 < cfields> It looks pretty clear that that's the correct guard to be using. 15:06 < cfields> I'd move the "#include " into that block. 15:07 < dongcarl> I feel like that's more prone to failure as I have no idea what `QT_BEGIN_NAMESPACE` does 15:10 < cfields> Ok 15:37 < cfields> dongcarl: I commented on the PR. My greps were failing me and I'm about to head out, so I'm afraid I left you a few hints rather than full suggestions. If any of that was unclear, feel free to ask. 15:37 < dongcarl> Cool. Thanks very much Cory! 15:37 < cfields> Great work though, this is a nice improvement. 15:37 < dongcarl> :-) 15:42 < dongcarl> Upstreamed... https://codereview.qt-project.org/c/qt/qtbase/+/268206 15:42 -!- jb55 [~jb55@gateway/tor-sasl/jb55] has quit [Remote host closed the connection] 15:42 < dongcarl> Not sure what their cherry-pick policy is, so we might need to maintain patch for a little while 17:39 < fanquake> Qts policy? 17:40 < fanquake> This is good progress. 17:40 < dongcarl> Yeah QT’s policy 21:11 -!- hebasto [~hebasto@95.164.65.194] has joined #bitcoin-builds 21:13 -!- hebasto [~hebasto@95.164.65.194] has quit [Remote host closed the connection] --- Log closed Fri Jul 19 00:00:15 2019