this time as tar.gz because adding (or downloading) 50+ files a bit of chore.. based on mainline c0251c4aa075 mostly same old patches for termux + few even older hacks.. Andrea, I broke my cpu_debug patch into 3. randrik9/0047-Debug-add-printing-of-cpucount-from-ff_cpus.patch randrik9/0048-Debug-add-cpu-count-printing-for-mwindow.C.patch randrik9/0049-Hack-raise-cpu-cap-from-8-to-16.patch can you (and Phyllis) see if just 0049 gives you better result? if possble please also test non-ffmpeg based MPEG encoding and decoding...
in theory, because I made all patches with git format-patch you can apply them all with 'git am *.patch' ... from top level dir where .git of cinelerra actuallyocated... On Monday, July 19, 2021, Andrea paz <gamberucci.andrea@gmail.com> wrote:
Can I put all the patches together? How do I patch the pkg-config?
Apply all patches with the following exceptions: 0014 and 0015 --> failed 0021 --> patching file libbthread.patch1 git binary diffs are not supported 0053 --> ? Should I create the directory .../cinelerra-5.1/my_configure? and put the patch in this directory? I'm waiting for some advice before compiling.
On Monday, July 19, 2021, Andrea paz <gamberucci.andrea@gmail.com> wrote:
Apply all patches with the following exceptions:
0014 and 0015 --> failed
hm, probably because files already differ from clean git tree this series was based on? Anyway they useful only for guarding x86-specific parts of code, for building on non-x86 randrik9/0014-ifdef-commercials-db-only-for-x86.patch randrik9/0015-fixed-x86-ifdefs.patch
0021 --> patching file libbthread.patch1 git binary diffs are not supported
hm, strange... need to look into it, but again this is termux-specific part and should be guarded by termux ifdefs, so not relevant on x86/Linux
0053 --> ? Should I create the directory .../cinelerra-5.1/my_configure? and put the patch in this directory?
no, i just added this script as way to remember specific long configure.. you can configure your usual way.. try to compile?
I'm waiting for some advice before compiling.
i re-did patches using git format-patch --full-index --binary as suggested here https://stackoverflow.com/questions/50677861/git-binary-diffs-are-not-suppor... attached On Tuesday, July 20, 2021, Andrew Randrianasulu <randrianasulu@gmail.com> wrote:
On Monday, July 19, 2021, Andrea paz <gamberucci.andrea@gmail.com> wrote:
Apply all patches with the following exceptions:
0014 and 0015 --> failed
hm, probably because files already differ from clean git tree this series was based on? Anyway they useful only for guarding x86-specific parts of code, for building on non-x86
randrik9/0014-ifdef-commercials-db-only-for-x86.patch randrik9/0015-fixed-x86-ifdefs.patch
0021 --> patching file libbthread.patch1 git binary diffs are not supported
hm, strange... need to look into it, but again this is termux-specific part and should be guarded by termux ifdefs, so not relevant on x86/Linux
0053 --> ? Should I create the directory .../cinelerra-5.1/my_configure? and put the patch in this directory?
no, i just added this script as way to remember specific long configure.. you can configure your usual way..
try to compile?
I'm waiting for some advice before compiling.
Without being root the compilation stops after a few seconds because it doesn't find the libzmpeg3 directory. As root the compilation goes on more, but also fails. I enclose the cin5-root.log Apologies for the Italian output; I tried using LC_ALL=C before to exec "make", but it still uses Italian... (NOTE: if it is possible, I would like to be able to continue compiling without becoming root)
On Tuesday, July 20, 2021, Andrea paz <gamberucci.andrea@gmail.com> wrote:
Without being root the compilation stops after a few seconds because it doesn't find the libzmpeg3 directory.
oh, probably artefact of termux enviroment... try chown your_system_name:your_group -R cinelerra/ also be sure to put ffmpeg-4.4 tarball and libaom-3.1.0 tarball into thirdparty/src (not included in patches due to size, try my cloud folder: https://cloud.mail.ru/public/M8Pz/M3xjrE7NV https://cloud.mail.ru/public/TLWC/Jr6dLMYFk )
As root the compilation goes on more, but also fails. I enclose the cin5-root.log Apologies for the Italian output; I tried using LC_ALL=C before to exec "make", but it still uses Italian... (NOTE: if it is possible, I would like to be able to continue compiling without becoming root)
@Phyllis I saw that in the last git you implemented some Andrew patches. Can I know what exactly they are so I can save myself something? I'm applying them one by one and it's taking me a lot of time.... @Andrew I can't download your ffmpeg and libaom tarballs. Can I download them from their websites? Do you have a direct address?
On Tuesday, July 20, 2021, Andrea paz <gamberucci.andrea@gmail.com> wrote:
@Phyllis I saw that in the last git you implemented some Andrew patches. Can I know what exactly they are so I can save myself something? I'm applying them one by one and it's taking me a lot of time....
I think you can apply ALL of them by saying 'git am path_to_dir_with_patches/*.patch' (so, * will be expanded to all patch names) but before you better to say 'git rest --hard first_few_symbols in commit you are reseting to' (be sure to backup all interesting changes in tree) and if any patchs fails to apply (because they already applied in main three) there hopefully will be helpful message saying you can use git am --skip..
@Andrew I can't download your ffmpeg and libaom tarballs. Can I download them from their websites? Do you have a direct address?
ow. try https://ffmpeg.org/download.html - > xz tarball for 4.4 and you can try to unpack src rpm from https://packages.altlinux.org/en/sisyphus/srpms/libaom/rpms but it will give you different directory structure compared to one assumed by cingg's buildsystem... there was official google source but ironically I can't download from it due to restrictions? https://storage.googleapis.com/aom-releases/libaom-3.1.1.tar.gz note you might want to repack this as xz .. and rename libaom patch accordingly. build a bit slow here so I might not have tested source by end of this day...
Like yesterday, today "git am" fails leaving many errors. See attachment. I guess, due to my incompetence, I can only apply the patches one by one. NOTE: having downloaded ffmpeg-4.4, should I delete ffmpeg-4.3 and all its patches?
On Tuesday, July 20, 2021, Andrea paz <gamberucci.andrea@gmail.com> wrote:
Like yesterday, today "git am" fails leaving many errors. See attachment. I guess, due to my incompetence, I can only apply the patches one by one.
hm, just repeat "git am --skip" as many times as needed?
NOTE: having downloaded ffmpeg-4.4, should I delete ffmpeg-4.3 and all its patches?
i think yes, unless you want to retest with it again?
I can't solve any of the problems. I am not able to do these tests. The following patches fail and I skipped them: 07-08-09-12-13-16-17-25-30 These others give errors (see attachment patch.txt): 35-36-37-40-50-51-53 Compilation is not occurring. See the other attachments. Couldn't you select just one set of patches and review those before moving on to others? @Phyllis The latest git, without further patching, works fine. One small note is about the location of BT601_PAL at the bottom between the 2 BT2020s. See image.
On Wednesday, July 21, 2021, Andrea paz <gamberucci.andrea@gmail.com> wrote:
I can't solve any of the problems. I am not able to do these tests. The following patches fail and I skipped them: 07-08-09-12-13-16-17-25-30
These others give errors (see attachment patch.txt): 35-36-37-40-50-51-53
did you reset your tree to latest official git commit first? I can regenerate patches on top of this again...
Compilation is not occurring. See the other attachments. Couldn't you select just one set of patches and review those before moving on to others?
pick for example x265-multilib...
@Phyllis The latest git, without further patching, works fine. One small note is about the location of BT601_PAL at the bottom between the 2 BT2020s. See image.
yeah, because I tried to keep numeric constants for previously-defined colorspaces the same, and order important here.. I'll try to reorganize things a bit... but not sure about outcome.
updated tarball attached. you probably still need git reset --hard 34d7b567ee523 first, then 'chown -R whole_tree' for yout user as for testing patches separately you can try randrik10/0034-Add-multilib-x265-slower-compilation-but-you-can-ren.patch randrik10/0035-Add-timecode-to-output-mov-mxf.patch randrik10/0036-Hack-pix_fmts-for-yuv4mpegpipe-muxer-encoder-allow-f.patch randrik10/0043-Hack-raise-cpu-cap-from-8-to-16-might-have-bad-effec.patch randrik10/0045-HACK-try-to-speed-up-reverse-playback-for-i-only-for.patch randrik10/0046-Add-openEXR-format-for-background-render.patch randrik10/0049-Add-x11-inc-autodetect-for-shuttle-target-using-pkg-.patch and additionally/after test randrik10/0039-Add-hidden-fast-slow-playback-preference.patch randrik10/0040-Fixup-titler-plugin-after-variable-fast-slow-speed-c.patch On Wednesday, July 21, 2021, Andrea paz <gamberucci.andrea@gmail.com> wrote:
I can't solve any of the problems. I am not able to do these tests. The following patches fail and I skipped them: 07-08-09-12-13-16-17-25-30
These others give errors (see attachment patch.txt): 35-36-37-40-50-51-53
Compilation is not occurring. See the other attachments. Couldn't you select just one set of patches and review those before moving on to others?
@Phyllis The latest git, without further patching, works fine. One small note is about the location of BT601_PAL at the bottom between the 2 BT2020s. See image.
updated tarball attached. you probably still need git reset --hard 34d7b567ee523
Each time I delete the CinGG folder (cinelerra5) and do a new download with git clone; I didn't think git reset was needed. After a git clone: [paz@arch-paz cinelerra-5.1]$ git checkout master Already in 'master' Your branch is updated with respect to 'origin/master'. [paz@arch-paz cinelerra-5.1]$ git reset --hard 34d7b567ee523 fatal: ambiguous '34d7b567ee523' argument: unknown revision or path not in working tree. Use '--' to separate paths from revisions, as follows: 'git <command> [<revision>...] -- [<file>...]'
first, then 'chown -R whole_tree' for yout user
Regarding folder access, I have tried several times to give: # chown -R paz:users /home/paz/cinelerra5 but this folder already belongs to paz, so nothing changes.
as for testing patches separately you can try:
I'm going to try just putting in the patches you pointed out, but I can't find randrik10.
On Wednesday, July 21, 2021, Andrea paz <gamberucci.andrea@gmail.com> wrote:
updated tarball attached. you probably still need git reset --hard 34d7b567ee523
Each time I delete the CinGG folder (cinelerra5) and do a new download with git clone; I didn't think git reset was needed. After a git clone:
[paz@arch-paz cinelerra-5.1]$ git checkout master Already in 'master' Your branch is updated with respect to 'origin/master'. [paz@arch-paz cinelerra-5.1]$ git reset --hard 34d7b567ee523 fatal: ambiguous '34d7b567ee523' argument: unknown revision or path not in working tree. Use '--' to separate paths from revisions, as follows: 'git <command> [<revision>...] -- [<file>...]'
hm, ok.... while strange.. you can just search for commit marked 'origin/master' and reset to that in future...
first, then 'chown -R whole_tree' for yout user
Regarding folder access, I have tried several times to give:
# chown -R paz:users /home/paz/cinelerra5
but this folder already belongs to paz, so nothing changes.
i assumed failure to build from user was due to my patches somewhat changing ownership of some files?
as for testing patches separately you can try:
I'm going to try just putting in the patches you pointed out, but I can't find randrik10.
mmmm... I tried to attach it ( another route is via cloud.mail.ru but this one was giving you troublesx...
I did other tests: - I used "git reset --hard"; I applied the patches with "git am", but I still get the "trailing whitespace" errors. I still tried compiling but it crashes after a few seconds. - I did a make clean; a "git reset --hard" and applied the patches manually, this time with no errors. I gave permissions back to the cinelerra5 folder, thinking that the patches might have altered something. I tried compiling, but again it crashed after a few seconds. Append the cin5.log.
On Wednesday, July 21, 2021, Andrea paz <gamberucci.andrea@gmail.com> wrote:
I did other tests: - I used "git reset --hard"; I applied the patches with "git am", but I still get the "trailing whitespace" errors. I still tried compiling but it crashes after a few seconds.
- I did a make clean; a "git reset --hard" and applied the patches manually, this time with no errors. I gave permissions back to the cinelerra5 folder, thinking that the patches might have altered something. I tried compiling, but again it crashed after a few seconds. Append the cin5.log.
tar -xJf src/ffmpeg-4.4.tar.xz CONFIGURING djbfft cd djbfft* && echo "/home/paz/cinelerra5/cinelerra-5.1/thirdparty/../thirdparty/djbfft-0.76/"
conf-home; (CFLAGS="-g -O2 '-DCIN="cin"' '-DCINLIB_DIR="IN_PATH"' '-DCINDAT_DIR="IN_PATH"' '-DCONFIG_DIR="/home/paz/.bcast6"' '-DNESTED_DIR="OME/Videos"' '-DSNAP_DIR="OME/Pictures"' '-DPLUGIN_DIR="IN_LIB/plugins"' '-DLOCALE_DIR="IN_LIB/locale"' '-DLADSPA_DIR="IN_LIB/ladspa"' '-DCIN_BROWSER="firefox"' -fno-omit-frame-pointer -fno-math-errno -fno-signed-zeros -pthread -Wall -Wno-unknown-warning-option -Wno-attributes -Wno-unused-result -Wno-stringop-overflow -Wno-format-truncation -Wno-format-overflow -Wno-parentheses -Wno-stringop-truncation -D__STDC_CONSTANT_MACROS -D__STDC_LIMIT_MACROS -DPNG_SKIP_SETJMP_CHECK=1 -D_LARGEFILE_SOURCE -D_LARGEFILE64_SOURCE -D_FILE_OFFSET_BITS=64 -I/usr/include -I/usr/local/include -I/usr/include/freetype2 -I/usr/local/include/freetype2 -I/usr/include/uuid -I/usr/local/include/uuid -I/usr/include/mjpegtools -I/usr/local/include/mjpegtools -DHAVE_CIN_3RDPARTY -DHAVE_GL -DHAVE_XFT -DHAVE_XXF86VM -DHAVE_OSS -DHAVE_ALSA -DHAVE_FIREWIRE -DHAVE_OGG -DHAVE_DV -DHAVE_DVB -DHAVE_LADSPA -DHAVE_VIDEO4LINUX2 -DHAVE_PULSE -DHAVE_PACTL -DHAVE_OPENEXR -DHAVE_LV2 -DHAVE_LIBZMPEG -DHAVE_SHUTTLE -DHAVE_SHUTTLE_USB -DHAVE_XV -DHAVE_VAAPI -DHAVE_VDPAU -DHAVE_NV -DHAVE_WINTV -DHAVE_X10TV -DBOOBY"; echo "cc $CFLAGS") > conf-cc; echo > ./configure; chmod +x ./configure; ./configure CONFIGURING twolame CONFIGURING openjpeg CONFIGURING lame cd twolame* && ./configure --enable-shared=no cd lame* && CFLAGS+=" -O" ./configure --enable-shared=no cd openjpeg* && echo 'cmake "$@" "."' > ./configure; chmod +x ./configure; ./configure -DBUILD_SHARED_LIBS:BOOL=OFF CONFIGURING opus cd opus* && ./configure CONFIGURING libogg xz: (stdin): File format not recognized CONFIGURING libvorbis tar: Child returned status 1 tar: Error is not recoverable: exiting now
==== VERY strange.... can you double-check your ffmpeg-4.4 tarball actually downloaded correctly, and normal tar can unpack it? ? also, this line and few others looks like it truncated at first symbol? -DCINLIB_DIR="IN_PATH"
I downloaded a new ffmpeg-4.4.tar.xz and now the compile error happened a little bit later. I attach the cin5.log. However I think that the fault is not the tar.xz but mine, that I had removed also the ffmpeg.git patches, during some tests, in addition to the ffmpeg-4.3 patches PS: I tried a build without patches, but including the 3 of h265 present in git. The installation works, but trying to render using h265 at 10bit and h265 at 12bit, leads to these errors: [libx265 @ 0x7f3fa57f2400] Specified pixel format yuv422p10le is invalid or not supported FFMPEG::open_encoder err: Invalid argument int FFMPEG::open_encoder(const char*, const char*): open failed libx265:/home/paz/video_editing/prova/10bit.mp4 Render::render_single: Session finished. [libx265 @ 0x7f3fa57ef400] Specified pixel format yuv422p12le is invalid or not supported FFMPEG::open_encoder err: Invalid argument int FFMPEG::open_encoder(const char*, const char*): open failed libx265:/home/paz/video_editing/prova/12bit.mp4 Render::render_single: Session finished.
On Wednesday, July 21, 2021, Andrea paz <gamberucci.andrea@gmail.com> wrote:
I downloaded a new ffmpeg-4.4.tar.xz and now the compile error happened a little bit later. I attach the cin5.log. However I think that the fault is not the tar.xz but mine, that I had removed also the ffmpeg.git patches, during some tests, in addition to the ffmpeg-4.3 patches
ffmoeg. git patches only should be relevant / active if you add - - with-git-ffmpeg=url to your configure....
PS: I tried a build without patches, but including the 3 of h265 present in git. The installation works, but trying to render using h265 at 10bit and h265 at 12bit, leads to these errors:
[libx265 @ 0x7f3fa57f2400] Specified pixel format yuv422p10le is invalid or not supported FFMPEG::open_encoder err: Invalid argument int FFMPEG::open_encoder(const char*, const char*): open failed libx265:/home/paz/video_editing/prova/10bit.mp4 Render::render_single: Session finished. [libx265 @ 0x7f3fa57ef400] Specified pixel format yuv422p12le is invalid or not supported FFMPEG::open_encoder err: Invalid argument int FFMPEG::open_encoder(const char*, const char*): open failed libx265:/home/paz/video_editing/prova/12bit.mp4 Render::render_single: Session finished.
can you check if resulting x265.a library (in thirdparty/x265-3.4) contains 10 and 12 bit symbols? just to be clear, you applied x265-multilib patch and it applied correctly (check with git apply -v, just git apply may not return error just silently skip patching)
On Wednesday, July 21, 2021, Andrea paz <gamberucci.andrea@gmail.com> wrote:
I downloaded a new ffmpeg-4.4.tar.xz and now the compile error happened a little bit later. I attach the cin5.log. However I think that the fault is not the tar.xz but mine, that I had removed also the ffmpeg.git patches, during some tests, in addition to the ffmpeg-4.3 patches
PS: I tried a build without patches, but including the 3 of h265 present in git. The installation works, but trying to render using h265 at 10bit and h265 at 12bit, leads to these errors:
chmod: impossibile accedere a './multilib.sh': File o directory non esistente /bin/bash: riga 1: ./configure: File o directory non esistente make[2]: *** [Makefile:358: /home/paz/cinelerra5/cinelerra-5.1/thirdparty/../thirdparty/build/x265.configure] Errore 127 make[2]: *** Attesa per i processi non terminati.... so patches somehow failed to apply fully.....
[libx265 @ 0x7f3fa57f2400] Specified pixel format yuv422p10le is invalid or not supported FFMPEG::open_encoder err: Invalid argument int FFMPEG::open_encoder(const char*, const char*): open failed libx265:/home/paz/video_editing/prova/10bit.mp4 Render::render_single: Session finished. [libx265 @ 0x7f3fa57ef400] Specified pixel format yuv422p12le is invalid or not supported FFMPEG::open_encoder err: Invalid argument int FFMPEG::open_encoder(const char*, const char*): open failed libx265:/home/paz/video_editing/prova/12bit.mp4 Render::render_single: Session finished.
Hi Andrew, What is the basis to get this to work? 1. Pull the current git. 2. Apply uyour patches. 3. Download and unpack ?? MatN On Wed, 21 Jul 2021 19:43:30 +0300 Andrew Randrianasulu via Cin <cin@lists.cinelerra-gg.org> wrote:
On Wednesday, July 21, 2021, Andrea paz <gamberucci.andrea@gmail.com> wrote:
I downloaded a new ffmpeg-4.4.tar.xz and now the compile error happened a little bit later. I attach the cin5.log. However I think that the fault is not the tar.xz but mine, that I had removed also the ffmpeg.git patches, during some tests, in addition to the ffmpeg-4.3 patches
PS: I tried a build without patches, but including the 3 of h265 present in git. The installation works, but trying to render using h265 at 10bit and h265 at 12bit, leads to these errors:
chmod: impossibile accedere a './multilib.sh': File o directory non esistente /bin/bash: riga 1: ./configure: File o directory non esistente make[2]: *** [Makefile:358: /home/paz/cinelerra5/cinelerra-5.1/thirdparty/../thirdparty/build/x265.configure] Errore 127 make[2]: *** Attesa per i processi non terminati....
so patches somehow failed to apply fully.....
[libx265 @ 0x7f3fa57f2400] Specified pixel format yuv422p10le is invalid or not supported FFMPEG::open_encoder err: Invalid argument int FFMPEG::open_encoder(const char*, const char*): open failed libx265:/home/paz/video_editing/prova/10bit.mp4 Render::render_single: Session finished. [libx265 @ 0x7f3fa57ef400] Specified pixel format yuv422p12le is invalid or not supported FFMPEG::open_encoder err: Invalid argument int FFMPEG::open_encoder(const char*, const char*): open failed libx265:/home/paz/video_editing/prova/12bit.mp4 Render::render_single: Session finished.
On Wednesday, July 21, 2021, mnieuw--- via Cin <cin@lists.cinelerra-gg.org> wrote:
Hi Andrew,
What is the basis to get this to work? 1. Pull the current git.
yes. 2. Apply uyour patches. yes (there are some whitespace errors, but they shouldn't prevent you from applying them..) 3. Download and unpack ?? if you want to test whole series download libaom 3.1.0/3.1.1 (xz repack) and ffmpeg-4.4.tar.gz (from their website) put them in thirdparty/src, try to build usually ... but you can just test single patches or small group of them...
MatN
On Wed, 21 Jul 2021 19:43:30 +0300 Andrew Randrianasulu via Cin <cin@lists.cinelerra-gg.org> wrote:
On Wednesday, July 21, 2021, Andrea paz <gamberucci.andrea@gmail.com> wrote:
I downloaded a new ffmpeg-4.4.tar.xz and now the compile error happened a little bit later. I attach the cin5.log. However I think that the fault is not the tar.xz but mine, that I had removed also the ffmpeg.git patches, during some tests, in addition to the ffmpeg-4.3 patches
PS: I tried a build without patches, but including the 3 of h265 present in git. The installation works, but trying to render using h265 at 10bit and h265 at 12bit, leads to these errors:
chmod: impossibile accedere a './multilib.sh': File o directory non esistente /bin/bash: riga 1: ./configure: File o directory non esistente make[2]: *** [Makefile:358: /home/paz/cinelerra5/cinelerra-5.1/thirdparty/../thirdparty/build/x265. configure] Errore 127 make[2]: *** Attesa per i processi non terminati....
so patches somehow failed to apply fully.....
[libx265 @ 0x7f3fa57f2400] Specified pixel format yuv422p10le is invalid or not supported FFMPEG::open_encoder err: Invalid argument int FFMPEG::open_encoder(const char*, const char*): open failed libx265:/home/paz/video_editing/prova/10bit.mp4 Render::render_single: Session finished. [libx265 @ 0x7f3fa57ef400] Specified pixel format yuv422p12le is invalid or not supported FFMPEG::open_encoder err: Invalid argument int FFMPEG::open_encoder(const char*, const char*): open failed libx265:/home/paz/video_editing/prova/12bit.mp4 Render::render_single: Session finished.
-- Cin mailing list Cin@lists.cinelerra-gg.org https://lists.cinelerra-gg.org/mailman/listinfo/cin
participants (3)
-
Andrea paz -
Andrew Randrianasulu -
mnieuw＠zap.a2000.nl