Dev Loading PNG data (or other) into a Raku variable instead of into a file.

[jforget]

This pull request adds function gdImagePngPtr (and similar gdImageXXXPtr functions).

Also: the POD documentation contains the list of libgd functions ported to GD::Raw.

[lizmat]

Hmm... the author tests don't pass:

% raku -I. xt/gdimagepngptr.rakutest
1..1
raku(5327,0x1f584e240) malloc: Double free of object 0x142805530
raku(5327,0x1f584e240) malloc: *** set a breakpoint in malloc_error_break to debug
zsh: abort      raku -I. xt/gdimagepngptr.rakutest

@jforget: could you verify / fix?

[jforget]

On my computer, the author tests run without any glitch. Yet, when I read the test script, I see that gdFree was called twice in a row, in lines 64 and 87, on both sides of my $test = ....

Fixed on my repo. Should I create a new pull request?

[lizmat]

No need for a new PR yet :-)

With that addition gdFree removed, the test runs ok, but fails:

% raku -I. xt/gdimagepngptr.rakutest
1..1
not ok 1 - 
# Failed test at xt/gdimagepngptr.rakutest line 89
# You failed 1 test of 1

[jforget]

I will be away from my keyboard for a few days, possibly until monday. Sorry about that.
In the meantime, can you store the content of variable $test in a permanent file, so I can compare it to ported-gdimagepngptr/gd-image-xxx-ptr.html?

Thank you in advance.

[lizmat]

https://gist.github.com/lizmat/bd4067d931a43fc4f94ddb8f95946a22

[jforget]

There is a tiny difference in the embedded JPEG data. This can be explained in a few ways.

• Your version of libjpeg is not the same as mine

• The JPEG generation uses a quality parameter and the default quality on your computer is not the same as on my computer

• JPEG is a lossy compression scheme, so there is some kind of unpredicatability (not sure about that reason, though).

Maybe I should at least remove testing gdImageJpegPtr from the test script.

Should I remove other formats? Maybe I should remove GIF? There are two functions for PNG: gdImagePngPtrEx with a required explicit $compression parameter and gdImagePngPtr which uses a system-wide value for the implicit compression. Should I discard gdImagePngPtr and keep only gdImagePngPtrEx?

[lizmat]

Perhaps a better test that wouldn't depend on identity of the whole?