ART now supports the usage of the lcms2-fast-float plugin, unfortunately the arch package of lcms2 does not include this plugin.
I've created the lcms2-ff-git AUR package that can be used with ART, but you need to explicitely install this package.
Git Clone URL: | https://aur.archlinux.org/art-rawconverter-git.git (read-only, click to copy) |
---|---|
Package Base: | art-rawconverter-git |
Description: | rawconverter ART (RawTherapee fork with ease of use in mind) built from latest sources |
Upstream URL: | https://bitbucket.org/agriggio/art/wiki/Home |
Keywords: | image photography raw |
Licenses: | GPL3 |
Conflicts: | art-rawconverter |
Provides: | art-rawconverter |
Submitter: | guzzisti |
Maintainer: | guzzisti |
Last Packager: | guzzisti |
Votes: | 2 |
Popularity: | 0.000311 |
First Submitted: | 2020-02-15 19:53 (UTC) |
Last Updated: | 2023-07-16 08:37 (UTC) |
ART now supports the usage of the lcms2-fast-float plugin, unfortunately the arch package of lcms2 does not include this plugin.
I've created the lcms2-ff-git AUR package that can be used with ART, but you need to explicitely install this package.
I added you as comaintainer of the -bin package.
ART when compiled from master uses only the commit checksum as version.
Do you mean that ART itself considers the version to be only the commit checksum? I was thinking for packaging purposes, it's useful to have some idea how newly built packages compare with releases.
We've used a similar version notation in the past, but upstream does not encourage to use such a notation. Actually ART when compiled from master uses only the commit checksum as version.
I can act as a co-maintainer for art-rawconverter-bin
if you like.
If you change pkgver()
to use:
git describe --long --tags | sed 's/\([^-]*-g\)/r\1/;s/-/./g'
The version numbers will be comparable with the normal release versions, like:
1.16.2.r14.g8d9f97933
Also, I created art-rawconverter-bin
for convenience because ART takes a very long time to compile. Would you be interested in taking it over or co-maintaining, since you're maintaining the other art-rawconverter packages?
I've reenabled the use of mimalloc
for memory allocation as per suggestion of upstream. Please report errors (especially memory related stuff like segfaults).
In a few days I'll gonna update the release package as well if everything is fine.
The new version enables build against libraw
to replace the vendored dcraw.c
with a more robust and better maintained raw decoding lib.
To enable the use of libraw
version 0.21 is required, so for the moment you need to have libraw-git
installed: https://aur.archlinux.org/packages/libraw-git/
@mervinb should be fixed now. Thanks for reporting the issue upstream.
I've been able to build the package, but I get the pink images on my Canon raw files. It's not 100% of the time - occasionally the same image appears correctly, the next time it's pink. Most of the time it's pink.
Looking at the makepkg log, I see this error, which might be useful:
[ 13%] Building CXX object rtengine/CMakeFiles/rtengine.dir/imagedata.cc.o
In file included from /usr/include/string.h:519,
from /usr/include/glib-2.0/glib/gslice.h:26,
from /usr/include/glib-2.0/glib.h:79,
from /home/mervin/.cache/yay/art-rawconverter-git/src/art-rawconverter-git_src/rtengine/dcraw.cc:12:
In function ‘void* memcpy(void*, const void*, size_t)’,
inlined from ‘int fread(void*, ssize_t, ssize_t, IMFILE*)’ at /home/mervin/.cache/yay/art-rawconverter-git/src/art-rawconverter-git_src/rtengine/myfile.h:123:16,
inlined from ‘void DCraw::parse_makernote(int, int)’ at /home/mervin/.cache/yay/art-rawconverter-git/src/art-rawconverter-git_src/rtengine/dcraw.cc:5643:13:
/usr/include/bits/string_fortified.h:34:33: warning: ‘void* __builtin___memcpy_chk(void*, const void*, long unsigned int, long unsigned int)’ specified size between 9223372036854775808 and 18446744073709551615 exceeds maximum object size 9223372036854775807 [-Wstringop-overflow=]
34 | return __builtin___memcpy_chk (__dest, __src, __len, __bos0 (__dest));
| ~~~~~~~~~~~~~~~~~~~~~~~^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
[ 14%] Building CXX object rtengine/CMakeFiles/rtengine.dir/imagedimensions.cc.o
FYI, when building rawtherapee-git, the same file is built without error, and the images appear correctly.
After the transition to gcc 10.1 images may become red. I've added a patch to adjust the build params to use -fno-tree-loop-vectorize
to mitigate this issue.
As the use of tcmalloc
leads to a segfault on startup (atleast on my system) i reverted this option for the moment.
Pinned Comments
guzzisti commented on 2023-01-15 09:33 (UTC)
ART now supports the usage of the lcms2-fast-float plugin, unfortunately the arch package of lcms2 does not include this plugin.
I've created the lcms2-ff-git AUR package that can be used with ART, but you need to explicitely install this package.