On 3/14/18, Antonio Scuri <antonio.sc...@gmail.com> wrote: > Tested on the following configurations: > > CentOS 6.8 (x86) Linux26g4c6 (GTK 2.24) > Debian 8.3 (x86) Linux316 (GTK 3.14) > Oracle Solaris 11 (x86) SunOS511x86 (GK 2.20) > Ubuntu 10.10 (x86) Linux26g4 (GK 2.22) > Ubuntu 12.04 (x86) Linux32 (GTK 2.24) > Ubuntu 14.04 (x64) Linux313_64 (GTK 3.10) > Ubuntu 15.04 (x64) Linux319_64 (GTK 3.14) > Ubuntu 16.04 (x64) Linux44_64 (GK 3.18) > > I run the test around 50 times. I added a small callback so I can do it > using only the keyboard. > > Using the IUP from SVN. > > The result was the same, every time... > > Best, > Scuri > > BTW: the SHRINK attribute you used along the code is a IupDialog attribute. > Other controls simply ignore it. Although I didn't change that code for the > tests. >
This is a very frustrating bug because sometimes I go very long times (days) without seeing it, and other times it always comes up. For the video recording, I probably had to run it 30+ times (and I edited out the other runs). It seems that whenever I want to see it, it never comes, but when I need it to work without problems (like a production app), it doesn't and I get long streaks of broken runs. Murphy's Law. I've seen it on the following using the standard packages that come with each OS: Ubuntu 12.04 LTS (x64), both GTK2 and GTK3 Ubuntu 14.04 (x64) (only tried GTK2) Ubuntu 16.04 (x64) (only tried GTK2) Ubuntu 16.10 (x64) (both GTK2, GTK3) Debian Jessie (x64) (GTK2) Debian Stretch (x64) (GTK2) SteamOS (x64) (GTK2) Arch this past month (x64) (only GTK2) Raspbian Jessie (armv7) (both GTK2, GTK3) Raspbian Stretch (armv7) (both GTK2, GTK3) I think I also got a report of the issue on Fedora Core x64 GTK2, but I don't have the specifics of which version. I originally was using the standard IUP SVN, tested the most on Ubuntu 12.04 and Raspbian Jessie. And I first reported this bug 2 years ago, and I've seen it over the entire duration of time. So I am almost certain this bug is in there. However, in recent days, I've been using a branch I made (called "Linux") from my IupCocoa repo. The branch doesn't actually contain the Cocoa stuff, but does include the CMake build system we've developed for all the platforms because I found the Makefile difficult to get going and use with IDEs. The branch also contains the event loop modification I described in the Lua Workshop video, but I did not activate it for the test runs and video recording, and I originally encountered this bug before I implemented that change. Anecdotally, I feel like I've seen the bug the most on Raspberry Pi (Raspbian). I don't know if it is scientifically true, but it feels like it. If it is true, I wonder if the slower speed of the Pi makes it more vulnerable since it feels like some kind of race condition bug. But if you want to try my branch, you can find it here. (Make sure to switch to the Linux branch.) But as I said, I don't believe my branch is affecting things since I did see it originally with the official SVN. I resync'd my branch with the SVN like a week and a half ago. https://github.com/ewmailing/IupCocoa/tree/Linux mkdir -p /path/to/some/build/dir cd /path/to/some/build/dir ccmake -DCMAKE_BUILD_TYPE=Debug -DIUP_BUILD_TESTS=1 -D /path/to/iup/root make There are menu options like for building tests (off by default which is the reason for the switch) which is off by default. I turn it on, and then hijack one of the tests (like button.c) to run the test code I submitted here. It attempts to auto-detect for GTK2 or GTK3, picking GTK2 first. Menu options can set to change that. (You can use ccmake or cmake-gui to bring up the menu.) Valid options for build type are Debug, MinSizeRel, Release, RelWithDebInfo. I usually use Debug and MinSizeRel, but I have seen this bug with all of them. The video recording was on Ubuntu 16.10 (x64) using GTK2. I built that using Debug settings. However, I originally encountered this bug with Release or MinSizeRel settings in a production app, so anecdotally, I wonder if the optimized variant is more easily able to hit the race condition. Anyway, I implore you to keep trying. I know the bug is there. Maybe building Release, trying a slower machine, or running a large and heavy background process to slow down the machine and eat resources could help trigger it. Thanks, Eric ------------------------------------------------------------------------------ Check out the vibrant tech community on one of the world's most engaging tech sites, Slashdot.org! http://sdm.link/slashdot _______________________________________________ Iup-users mailing list Iup-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/iup-users
