Newer Apple devices are creating EXIF without this however libexif
still requires it due to JPE...

* seems to export to cmyk+2 correctly

We'll need something fancier for import.

* more hack...

* CI: upgrade macOS runner to version 12 (Monterey)

* CI: upgrade actions/checkout to v3

*...

* spngsave: ensure quantisation occurs last

i.e. after setting the metadata and phyiscal pixe...

* webpsave: switch to g_try_malloc()

* webpsave: add dimensions limit check

* output the a...

now parallel 16-bit PNG save

see https://github.com/libvips/libvips/discussions/3087

* vala: Enhance the metadata to contain more symbols

* Enhance the introspection coverage

...

* simple invert example

* Update meson.build

The PNG spec requires that the iCCP chunk has a profile name with
a minumum length of 1 (we use...

Since that could terminate the application in debug builds. Also,
add the missing case clause f...

Also, remove a remnant of Autotools.

Instead of our own compiler attr thing.

see https://github.com/libvips/libvips/issues/2871

We were only testing "killed" once for each strip of tiles. Check on each
tile as well to make s...

We were not caching seq mode loaders used in random access mode.

see https://github.com/libvips...

* reducev: initialize the sum with the addition

Avoids an Orc opcode.

* convi: initialize ...

* reducev: use convsuswb in Orc path

Saves a few instructions.

* convi: use convsuswb in O...

- fix sign issue with some chunk sizes
- allow larger chunk_len

See https://github.com/libvips/...

Ensures the quality to distance conversion is continuous at 30.

See:
https://github.com/libj...

* Inverted order of TIFF tag writing when saving multi-page subifd pyramids to ensure that subif...

We added a check that loader ->header() and ->load() methods set the
same dhint. This commit rem...

Some libvips header fields can be NULL, for example filename, and we
need to avoid null pointer ...

Some libvips header fields can be NULL, for example filename, and we
need to avoid null pointer ...

adding the examples/ if/endif broke the sed that's used to remove extra
subdirs in ci

plus one example

* webpsave: switch to sink_disc

* fix some warnings

* distinguish between animated and sin...

a bit simpler too

copying dloebl's idea from https://github.com/libvips/libvips/pull/3018

Resolves: #3024.

We could return an operation from cache even after the operatuion had
been blocked. This was har...

The GLib utilities, which was split into its own `glib-utils`
formula has been merged back into...

Add a check that header and load methods agree on the demand hint, and
make sure all loaders pas...

* move tiff decompress outside lock

Most time in TIFF read is spent in decompression. If we m...

* Implement `GsfOutfileStdio` by ourselves

* Add missing `->root` init

* Fix styling

* ...

see https://github.com/libvips/libvips/issues/3016

see https://github.com/libvips/libvips/discussions/3011

see https://github.com/libvips/libvips/discussions/3011

to make it big enough for an A4 page at 300dpi

see https://github.com/libvips/libvips/pull/2984

* cgifsave: add support for interlaced GIF write

note: cgif >= v0.3.0 is required

* switch...

Looks like there's no way to reset the decoder state, so we have to
remake it.

* jxlsave: avoid using deprecated functions

The upcoming libjxl 0.7 has deprecated a number o...

It's required to close the input, otherwise the encoder can't
know what the last frame is, resu...

SOme combinations of high bitdepth images with low bitdepth PNG save could
produce incorrect ima...

There was a missing shift, thanks jeffska

See https://github.com/libvips/libvips/issues/2985

Confusingly we previously saved jp2 images as a jpeg2000 codestream rather
than as a jp2 file.

...

Having a different colorspace and chroma subsampling just because the
source image happens to b...

Hopefully to make it clearer and more consistent.

See https://github.com/libvips/libvips/discus...

We were avoiding /0 by testing for alpha==0, however, this will still
allow very small values of...

We were minimising at the end of threadpool_run, but this was being
called for each sinkscreen r...