I tested with a sample SVG:

$ vips svgload lion.svg x.v --rgb128

And I get:

Very nice!

... and thank you for doing this work, Konrad, good job.

Thanks for your review @jcupitt . I did my best to apply all your great suggestions. Please take a look again at your convenience.

Commit kleisauke@c4d898f has some minor formatting improvements and fixes the compilation when HAVE_128BIT_SVG_RENDERING is undefined, e.g.:

../libvips/foreign/svgload.c: In function ‘vips_foreign_load_svg_load’:
../libvips/foreign/svgload.c:707:13: error: ‘svg’ undeclared (first use in this function)
  707 |         if (svg->rgb128) {
      |             ^~~
../libvips/foreign/svgload.c:707:13: note: each undeclared identifier is reported only once for each function it appears in
../libvips/foreign/svgload.c:708:28: error: ‘class’ undeclared (first use in this function)
  708 |                 vips_error(class->nickname, "128-bit rendering is not supported with this version");
      |                            ^~~~~

Feel free to cherry-pick.

Commit kleisauke@c4d898f has some minor formatting improvements and fixes the compilation when HAVE_128BIT_SVG_RENDERING is undefined, e.g.:

../libvips/foreign/svgload.c: In function ‘vips_foreign_load_svg_load’:
../libvips/foreign/svgload.c:707:13: error: ‘svg’ undeclared (first use in this function)
  707 |         if (svg->rgb128) {
      |             ^~~
../libvips/foreign/svgload.c:707:13: note: each undeclared identifier is reported only once for each function it appears in
../libvips/foreign/svgload.c:708:28: error: ‘class’ undeclared (first use in this function)
  708 |                 vips_error(class->nickname, "128-bit rendering is not supported with this version");
      |                            ^~~~~

Feel free to cherry-pick.

Thanks @kleisauke ! I cherry-picked whole your commit.

Great! It's looking rather neat now, I think.

We need two more things:

• doc comments near the bottom of svgload.c: you need to add rgb128 to the list of optional arguments for vips_svgload*(), and vips_svgload() needs a paragraph explaining what the flag does, eg. "Setting @rgb128 TRUE enables 128-bit scRGB output."
• this is an API change, so we need a line in the main CHANGELOG
• don't forget to credit yourself!

Sorry to bring this up so late, but I wonder if rgb128 is the best name for this flag? It's RGBA, not RGB, and it's 128 bits per pixel not per band element (the size measure we use elsewhere).

Savers use bitdepth to control the write depth of images. Maybe we should have a bitdepth parameter which defaults to 8 but can be set to 32? Can anyone think of another loader with a similar setting?

Does anyone have any thoughts on this?

Sorry to bring this up so late, but I wonder if rgb128 is the best name for this flag? It's RGBA, not RGB, and it's 128 bits per pixel not per band element (the size measure we use elsewhere).

Savers use bitdepth to control the write depth of images. Maybe we should have a bitdepth parameter which defaults to 8 but can be set to 32? Can anyone think of another loader with a similar setting?

Does anyone have any thoughts on this?

That's a very good point. I was also thinking about it for a while. I did some quick research and haven't found any loaders exposing such option. On one hand - bitdepth seems to follow naming convention touching such aspects and following it seems natural. On the other though - if we have only two options (8 and 32) for now (and probably this won't change soon) keeping it as bool is also a bit less misleading to me, this might be all about the parameter name we should change, perhaps something like: high_bit_depth flag with documented 32-bit value?

I'm happy to refactor it to be bitdepthanyway if that's what you guys prefer.

Great! It's looking rather neat now, I think.

We need two more things:

* doc comments near the bottom of `svgload.c`: you need to add `rgb128` to the list of optional arguments for `vips_svgload*()`, and `vips_svgload()` needs a paragraph explaining what the flag does, eg. "Setting @rgb128 TRUE enables 128-bit scRGB output."

* this is an API change, so we need a line in the main CHANGELOG

* don't forget to credit yourself!

Thanks! I'd postpone applying this for a while till we have a conclusion re: rgb128 :)

Great! It's looking rather neat now, I think.

We need two more things:

* doc comments near the bottom of `svgload.c`: you need to add `rgb128` to the list of optional arguments for `vips_svgload*()`, and `vips_svgload()` needs a paragraph explaining what the flag does, eg. "Setting @rgb128 TRUE enables 128-bit scRGB output."

* this is an API change, so we need a line in the main CHANGELOG

* don't forget to credit yourself!

@jcupitt I've followed your instructions with an assumption for now that keeping a bool flag, but renaming it to high_bitdepth and well-documented logic that's behind it would be the way to go.

OK, merged! Thank you for working on this feature, @kstanikviacbs.

This wouldn't be possible without your assistance @jcupitt . Thanks for that and making it possible to introduce it.