- Cin - Cinelerra-GG Mailing Lists

Mailing List
by Andrea paz 17 Dec '25

17 Dec '25

I usually read the mailing list starting from the archive so that I always have all the monthly threads at a glance. It seems to me that since it was reopened, the archive does not show the most recent messages, continuing to show only the first three emails sent. Am I doing something wrong? Is the archive updated only monthly or at other intervals? The mailing list works, because I always see all emails as soon as they arrive in my email client. I am only referring to how the archive works. It is not even important, because I imagine I am the only one who views the mailing list from the archive; I am simply curious. The archive address I use is the following (it may not be the correct one): https://lists.cinelerra-gg.org/archives/list/[email protected]/

4 8

Forum
by Andrea paz 17 Dec '25

17 Dec '25

For a few days now, the layout of the “artist forum” has been confusing. Is this happening to you too?

3 3

-Wcast-align warnings
by Andrew Randrianasulu 17 Dec '25

17 Dec '25

Trying to see if there were more such warnings than simply in guicast/bootstrap.C (this one kills compiled program on Sparc32) I added CFLAGS += -Wcast-align to global_config after configure generated it Aaaand ... it seems that our python code really generates c files with loads of those cast-align warnings, but some core files have them too ... I honestly do not know how to properly fix this :( v make -j8 > MLOG 2>&1 grep warning MLOG | grep Wcast | wc -l 1921 bccapture.C:77:24: warning: cast from 'const char *' to 'const u_int32_t *' (aka 'const unsigned int *') increases required alignment from 1 to 4 [-Wcast-align] bcresources.C:417:20: warning: cast from 'const char *' to 'const u_int32_t *' (aka 'const unsigned int *') increases required alignment from 1 to 4 [-Wcast-align] bcresources.C:1842:19: warning: cast from 'char *' to 'uint32_t *' (aka 'unsigned int *') increases required alignment from 1 to 4 [-Wcast-align] bcresources.C:1843:19: warning: cast from 'char *' to 'uint32_t *' (aka 'unsigned int *') increases required alignment from 1 to 4 [-Wcast-align] bcresources.C:1850:19: warning: cast from 'char *' to 'uint32_t *' (aka 'unsigned int *') increases required alignment from 1 to 4 [-Wcast-align] bcresources.C:1851:19: warning: cast from 'char *' to 'uint32_t *' (aka 'unsigned int *') increases required alignment from 1 to 4 [-Wcast-align] bcwindowbase.C:475:25: warning: cast from 'const char *' to 'const u_int32_t *' (aka 'const unsigned int *') increases required alignment from 1 to 4 [-Wcast-align] units.C:584:7: warning: cast from 'const char *' to 'const uint32_t *' (aka 'const unsigned int *') increases required alignment from 1 to 4 [-Wcast-align] units.C:617:7: warning: cast from 'const char *' to 'const uint32_t *' (aka 'const unsigned int *') increases required alignment from 1 to 4 [-Wcast-align] bcwindowbase.C:4448:18: warning: cast from 'uint8_t *' (aka 'unsigned char *') to 'unsigned int *' increases required alignment from 1 to 4 [-Wcast-align] vframe.C:968:15: warning: cast from 'unsigned char *' to 'float *' increases required alignment from 1 to 4 [-Wcast-align] vframe.C:971:15: warning: cast from 'unsigned char *' to 'float *' increases required alignment from 1 to 4 [-Wcast-align] vframe.C:973:15: warning: cast from 'unsigned char *' to 'float *' increases required alignment from 1 to 4 [-Wcast-align] vframe.C:975:15: warning: cast from 'unsigned char *' to 'float *' increases required alignment from 1 to 4 [-Wcast-align] vframe.C:993:3: warning: cast from 'unsigned char *' to 'uint16_t *' (aka 'unsigned short *') increases required alignment from 1 to 2 [-Wcast-align] ==== and so on ... They are harmless on x86/arm, but surely kills program on mips64/sparc (and may be Alpha).

1 1

Make x265 multidepth build configurable
by Andrew Randrianasulu 16 Dec '25

16 Dec '25

using old switch --enable-x265_hidepth build x265 10/12bit ([yes]) Note, i think I'll rebase this on top of Vulkan/etc patches but at some point Phyllis asked if there was way to make x265 build only 8bit variant for faster testing. I think now it is!

3 11

New patches
by Andrea paz 13 Dec '25

13 Dec '25

I have recently tested the following patches: 0001-Add-Vulkan-detection-to-buildsystem.patch 0004-External-libzimg-support-for-buildsystem.patch 0002-Add-libplacebo-Vulkan-filter-detection.patch 0019-Set-postprocessing-in-libtheora-decoder.patch 0002-Make-x265_hidepth-actually-work.patch 0023-Fixup-seg-name-to-segment-so-format-gui-button-work.patch 0003-Fix-libplacebo-enabling-was-typoed-wrong.patch Other patches are already in git, I'm not sure which ones still need to be tested. I need number 0004 (libzimg), otherwise I can't compile. The others work fine. I don't quite understand what I should test with x265. Another thing, I wanted to create encoding presets with AMD's AMF framework (the equivalent of Nvidia's NVENC), since it's open source and has been separated from amdgpu-pro, but I can't enable it in the ./configure phase. Any idea how to do this? One last thing, not important, just something I'm curious about: When I load an av1 media file, I get this warning: "Decoder libdav1d does not support device type vulkan. HW device init failed, using SW decode" I wanted to try disabling dav1d and going back to the classic libaom_av1 to test whether hardware-accelerated Vulkan had reached or exceeded the performance of dav1d (which is very fast but software-based). I can't disable dav1d; I tried editing the “decode.opts” file by commenting out the line: remap_video_decoder libaom-av1=libdav1d Or by replacing libdav1d with av1_vulkan; but I always get the same warning indicating that dav1d is still running. I also tried disabling dav1d in the ./configure phase (--disable-libdav1d) but it is not taken into account and I continue to get the same warning. Do you know how I could do this? (I remind you that this is just my curiosity, it is not important).

3 8

Possibly unneeded libs in thirdparty
by Andrew Randrianasulu 10 Dec '25

10 Dec '25

We grow a bit more so I looked if there was any lost old libs in thirdparty ... I think I found few! flac-1.4.3.tar.xz (currently at flac 1.5.0) openjpeg-2.5.0.tar.xz (at 2.5.2) tiff-4.6.0.tar.xz at 4.6.0t (at 4.7.0) ffmpeg 7.0 (currently at 8.0) x265_3.5.tar.xz (at 4.0 ) x264-snapshot-20210615-master.tar.xz and x264-stable.tar.gz from 2022 (currently at x264-stableR3191.tar.xz from 2024)

2 1

patches for djbfft - potentially fix another packaging error on Debian 13
by Andrew Randrianasulu 10 Dec '25

10 Dec '25

Apparently Debain 13's packaging is more strict, so some warnings turned into errors. I tried to adapt patch from Gentoo bugtracker, but not tested it beyond compilation with gcc 11.2.0 I guess test for this will be playing ac3 audio with cingg's native decoder (not ffmpeg)

3 4

Add few lines about debian pkg build to README
by Andrew Randrianasulu 09 Dec '25

09 Dec '25

But then I guess we need to keep changelog entries in debian/ dir and PKGBUILD and spec up to date so it will say current release date and not stuck at 2020. I'll try to fix Rosa spec (are new download locations download.cinlerra-gg.org stable? I guess /downloads will not come back ..)

1 0

Remove linux-firmware from debian/control file, does not exist in Deb 13
by Andrew Randrianasulu 09 Dec '25

09 Dec '25

I think we build mostly from qemu VMs where this file not needed, and open code builds even without X server actually using dri . Another use of linux-firmware is wifi firmware, but if you get network up enough for getting cin from net (or build on isolated machine) - it also not needed.

1 0

Debian 13.2.0 cinelerra-gg shared libs apt list
by Andrew Randrianasulu 08 Dec '25

08 Dec '25

in addition to packages installed by blds/bld_prepare.sh I installed apt install git apt install mc rsync just convenience for me ===== apt install lv2-dev apt install libsratom-dev apt install libsuil-dev apt install liblilv-dev === ffmpeg 6, 7 or 8 should do === apt install libavcodec-dev apt install libavformat-dev apt install libavfilter-dev ====== apt install liba52-dev apt install libtwolame-dev apt install libdv-dev apt install libmp3lame-dev apt install libavc1394-dev apt install libiec61883-dev May be we should add them to new script ? (shared builds were not a thing back in the day ..) Otherwise it now builds with --with-single-user --without-thirdparty --without-libdpx but on 2-way SMP virtual machine (qemu 9.0.something), with 2 Gb of ram. May be more cores and more ram expose some race condition in thirdparty build ?

3 9