tde-devs@chat.jabb.im < 2021/08/24 >
[00:58]michelec2 has joined
[02:14]michelec2 has left
[02:14]michelec2 has joined
[03:29]michelec2 has left
[08:50]Slávek has left
[08:52]cethyel has joined
[09:52]Slávek has joined
[09:56]cethyel: Slávek you're here that is good, I'm stopping writing my comments on aKode to let you know them ,here that I get FTBFS with ffmpeg-3.2.15 ( the one provided in Debian/Stretch) but It does build with ffmpeg-3.4.8 (Ubuntu/Bionic) :)
[09:57]Slávek: Thank you for the warning! I will explore it later during the day.
[09:57]cethyel: not sure if you intend to package R14.0.11 for Stretch though
[09:59]cethyel: only few warnings in ffmpeg-3.4.8 and here is the error I get with ffmpeg-3.2.15:
[09:59]cethyel:
/home/cethyel/CMAKE_CONVERSION/akode/akode/plugins/ffmpeg_decoder/ffmpeg_decoder.cpp: In member function ‘void aKode::FFMPEGDecoder::closeFile()’:
/home/cethyel/CMAKE_CONVERSION/akode/akode/plugins/ffmpeg_decoder/ffmpeg_decoder.cpp:235:9: error: ‘avio_context_free’ was not declared in this scope
235 | avio_context_free(&d->stream);
[11:59]michelec2 has joined
[12:03]michelec2: stretch will definitley be supported for quite a while
[12:21]cethyel: Hey Michele :)
[12:26]michelec2: Hi Greg
[12:40]Slávek: I seem to find the cause of FTBFS on Stretch... see comment on TDE/akode#7
[12:40]tde-bot: [TGW][akode] #7 - Revive FFmpeg Plugin
https://mirror.git.trinitydesktop.org/gitea/TDE/akode/pulls/7
[12:59]cethyel: I'm looking to adding the PATH to "rpc/rpc.h" in tqt3 (/usr/include/tirpc) because It seems that this header was removed from glibc, as a result packagers have been adding the tirpc package as a dependency and they PATH manually in their packages. So I adjusted the "nis.test" for that; It seems that I have to fill in the $QMAKE_VARS with a new INCLUDEPATH+ set on /usr/include/tirpc in the "configure" script, but I have to read the other shell scripts since I don't know exactly how to do It. Unfortunately I'm so lazy right now that I can't bring myself to the task Damned! >:-(
[13:14]cethyel: @Slávek, setting AV_VERSION_INT to 57, 80, 100 does work, the build succeed. btw, one may note that CMakeLists.txt can be amended to build the ffmpeg decoder along with the other modules.
[13:15]Slávek: Turn it ON by default? It seems as a good idea.
[13:15]cethyel: WITH_ALL_OPTIONS
[13:15]Slávek: (Y)
[13:16]cethyel: I have no ideas how to test if that decoder works actually
[13:18]cethyel: Amarok has an aKode backend I believe (from my red fish memory that lasts 10s...)
[13:19]michelec2: it has, if my memory is as good as yours :-)
[13:19]cethyel: if i build aKode with only ffmpeg then set Amarok on this backend, could be a valid test
[13:19]Slávek: Yes it is true for Amarok. As far as I know arts should also use akode if it is available.
[13:24]cethyel: well i'll be having something to eat and set myself to do something...tomorrow ;-) guys, see you!
[13:24]cethyel has left
[13:24]michelec2: see you Greg
[14:12]Slávek: Michele, you want to pay efforts for OpenMP with Clang in libkdcraw? Or for now we leave it enabled only for GCC?
[14:16]michelec2: I can give it a test, no problem
[14:16]michelec2: I haven't followed up tbh. What is required? testing? or debugging something not working?
[14:16]Slávek: It causes FTBFS.
[14:17]michelec2: oh ok.
[14:18]Slávek: The primary problem seems in "#pragma omp" definitions here:
https://mirror.git.trinitydesktop.org/gitea/TDE/libkdcraw/src/branch/master/libkdcraw/libraw/libraw/libraw_types.h
[14:18]michelec2: what if we remove the pragma directive?
[14:19]Slávek: This is only a structure definition, but it contains #pragma omp definitions. However, these do not contain any code only define "firstprivate" and "shared".
[14:19]michelec2: ok.
[14:20]michelec2: I will have a look later this week, maybe after tomorrow
[14:20]Slávek: When #pragma omp removes, it built fine.
[14:20]michelec2: does it give SEGV when testing it?
[14:20]Slávek: I didn't perform tests. Only build with GCC and Clang.
[14:21]michelec2: ok. what is a good way to test once I build with clang?
[14:22]Slávek: I do not know if there may be useful test programs built with the library.
[14:23]michelec2: ok, no worries.
[14:24]michelec2: I will have a look after I am done with some work I am doing on uncrustify
[14:24]michelec2: btw here some useful docs about omp pragmas
https://docs.microsoft.com/en-us/cpp/parallel/openmp/reference/openmp-directives?view=msvc-160
[14:25]michelec2: it seems those directive may be related tio performance optimization
[14:25]michelec2: at first sight at least
[14:25]Slávek: https://www.openmp.org/wp-content/uploads/OpenMP-API-Specification-5-1.pdf
[14:25]tde-bot: Page title:
[14:26]Slávek: When I looked at it, there seems to be no pragma variant that is not related to code block - only with the definition as is used in libkdcraw.
[14:28]michelec2: ok. let me check and we can discuss again in a couple of days
[14:29]Slávek: Basically, we can, for the time being, merge a variant with limit to GCC and to resolve newer versions of OpenMP to leave for later.
[14:30]michelec2: ok for me
[14:30]Slávek: However, if tests would show that sufficient is to remove #pragma omp from the header file and everything will be fine, it would also be a good solution :)
[14:32]michelec2: yes, of course. we need to understand what impact it would have on performances.
[14:32]michelec2: if any
[15:06]michelec2: time for bed for me
[15:06]michelec2: see you tomorrow and have a good evenign
[15:06]Slávek: good night to you
[15:06]michelec2: thanks
[15:06]Slávek: thanks
[15:06]michelec2 has left
[16:50]Slávek has left
[18:17]Slávek has joined
[22:17]Slávek has left
[22:47]Slávek has joined

tde-devs@chat.jabb.im < 2021/08/24 >