This will need to be re-targeted at the staging branch.

So I rebase and force push onto staging like normally, or is there another way? Because last time I did that, it pulled in a million different people via mention and I want to avoid that.

I'm told the trick is documented in #166472

It has worked. Thank you.

I have updated this to now also support SVG and the HEIF image format. Both of which are recent additions to imlib2.

Changelog: https://git.enlightenment.org/legacy/imlib2.git/tree/ChangeLog?h=v1.8.1

Also mentioned is LZMA loader.

The LZMA Loader is already active

------------------------------------------------------------------------
imlib2 1.8.1
------------------------------------------------------------------------
Configuration Options Summary:
Image loaders:

Regular image loaders

GIF.....................: yes

HEIF....................: yes

JPEG....................: yes

PNG.....................: yes

SVG.....................: yes

TIFF....................: yes

WEBP....................: yes

Decompressors

BZIP2...................: yes

LZMA(XZ)................: yes

ZLIB....................: yes

Containers

ID3.....................: yes
Build for X11.............: yes

Use X MIT-SHM FD-passing..: yes
Use X86 MMX for speed.....: no

Use AMD64 for speed.......: yes
Use visibility hiding.....: yes
Installation path.........: /nix/store/by1584shbcr2vwmlz17mj0lraaapb8af-imlib2-1.8.1

On macos 10.15:

  CCLD     bz2.la
ld: unknown option: -z
clang-11: error: linker command failed with exit code 1 (use -v to see invocation)
make[4]: *** [Makefile:741: heif.la] Error 1
make[4]: *** Waiting for unfinished jobs....
  CCLD     lzma.la
ld: unknown option: -z
clang-11: error: linker command failed with exit code 1 (use -v to see invocation)
make[4]: *** [Makefile:765: svg.la] Error 1
make[4]: Leaving directory '/private/tmp/nix-build-imlib2-1.8.1.drv-0/imlib2-1.8.1/src/modules/loaders'
make[3]: *** [Makefile:401: all-recursive] Error 1
make[3]: Leaving directory '/private/tmp/nix-build-imlib2-1.8.1.drv-0/imlib2-1.8.1/src/modules'
make[2]: *** [Makefile:401: all-recursive] Error 1
make[2]: Leaving directory '/private/tmp/nix-build-imlib2-1.8.1.drv-0/imlib2-1.8.1/src'
make[1]: *** [Makefile:523: all-recursive] Error 1
make[1]: Leaving directory '/private/tmp/nix-build-imlib2-1.8.1.drv-0/imlib2-1.8.1'
make: *** [Makefile:430: all] Error 2
builder for '/nix/store/9h4qw6713796w2jk1adj4m7qqm8f0b27-imlib2-1.8.1.drv' failed with exit code 2

Edit: this doesn't occur with the HEIF/SVG commit reverted

Think this is probably caused by the line https://git.enlightenment.org/legacy/imlib2.git/tree/src/modules/loaders/Makefile.am?h=v1.8.1#n110. Darwin clang apparently doesn't support this. So we either patch it out (not fully understanding why it's there) or disable svg on macos.

Aha https://git.enlightenment.org/legacy/imlib2.git/commit/src/modules/loaders/Makefile.am?h=v1.8.1&id=ffbd8bbca06f5ea608b8f78e73148515adfa8ecd

Should I take out the SVG and HEIF commit? I cannot test on Darwin.

No I'd simply disable the svg support if stdenv.isDarwin for now. Preferably with a comment explaining why.

We should probably notify upstream of this, but their phabricator instance https://phab.enlightenment.org/ is a bit annoying about getting an account.

I have updated it. Lets see if it is right now.

@ofborg eval

Ofborg seems to fail because it cannot evaluate passthru.tests on Darwin. I don't understand, there are no passthru tests defined.

#167079 Yes there are 😁

It's evaluating the result merged with staging.

Honestly I don't usually expect staging PRs to pass ofborg completely because of the huge volume of rebuilds that can be involved. Though it's usually worth taking a look at them for glaring errors anyway.

I do wonder if we should maybe be using lib.availableOn in some way for passthru.tests.

Hold on... that's not merged yet. Which error were you referring to?

From the Ofborg evaluation in this box down below with a click on "Details". https://github.com/NixOS/nixpkgs/pull/166452/checks?check_run_id=5893102214 . It says:

Unexpected error: command failed with exit code 1

Attempted: imlib2

The following builds were skipped because they don't evaluate on aarch64-darwin: imlib2.passthru.tests

Yep I think that's just staging being staging. I think the error message is a bit wrong because it really fails due to its build of many dependencies timing out.

Let's try and get this in to 22.05

It does not build on *-darwin:

• https://hydra.nixos.org/build/175483955
• https://hydra.nixos.org/build/175573398

Perhaps the workaround should've been tested (e.g. atop master) before merging.

Yeah teach me to assume something not tested for 24 days would stay the same.

Guessing one of the existing dependencies now pulls in librsvg itself. This is why we should use explicit "enable" flags when we can identify them.

For an added treat, upstream's git appears to have just fallen off the face of the earth too.

Version 1.9.0 is out already, which may address the issue. I see some mentionings of the svg loader in the changelog. I can make a PR for that as well.

https://sourceforge.net/projects/enlightenment/files/imlib2-src/1.9.0/

I've already tried this and it doesn't help.

Hold on - I think this may be the HEIF support too...

Think I've got this fixed in #171947

@nagy by all means create a 1.9.0 PR though, got to update sometime