* libnsgif: update script: Fix to handle dir with no patches
* libnsgif: Update to latest upstream
Fixes loading of broken gifs with truncated final frame.
* nsgifload: Call nsgif_data_complete after data scan
This allows libnsgif to distinguish between awaiting more
data, and a broken truncated GIF. In the latter case we
can display what we have.
* reimplement threadpool
just a set of threads that get recycled
"ninja test" passes and dzsave seems to work, though pytest fails for
some reason
* clean up threading code a bit
move base stuff into thread.c, so we have a simple
thread -> threadset -> threadpool
layering
* start trying to revert g_threadpool
based on the original commit
* working!
nice low systime again
still need to repply cnages to threadpool.c since 80e0cc3d1
* reapply fixes from master
so threadpool.c is now up to date
* rename VipsThread as VipsWorker
a bit less confusing
* use a semaphore to count workers in a pool
* tidy up
* formatting
* dynamic threadpool sizing
based on counting the number of blocked threads in each pool
it works, but the improvement is not great :(
* add "concurrency" metadata item
so operators can hint threadpool size (dzsave especially)
* don't use thinstrip for small images
* add RGB mode to openslide
since flatten was taking 20% of CPU time for dzsave
* fix up rgb mode
now actually works
* make the tile buffer per thread
in the new openslideload rgb mode
* fix dynamic pool downsize
* mild refactoring
* fix the buffer system
oops, turned it off by mistake
* all done!
* revise changelog
* Update libvips/iofuncs/threadset.c
Co-authored-by: Kleis Auke Wolthuizen <github@kleisauke.nl>
* Update libvips/iofuncs/threadset.c
Co-authored-by: Kleis Auke Wolthuizen <github@kleisauke.nl>
* LSan: add libMagickCore to suppression file
* Revert "Remove mutex lock for VipsThreadStartFn"
This reverts commit 41440491.
* add VIPS_MAX_THREADS
to set a hard limit on the threadset size
* Revert "Revert "Remove mutex lock for VipsThreadStartFn""
This reverts commit 77e8520966ba79194fff3b4e648bbd295cd5c260.
* remove sslock from sink.c
* move fixed threadpool build to init
not first use
* add some doc comments
* revert test suite threshold change
* add a test for MAX_THREADS
and move the test tmp/ area into the builddir
* limit VIPS_MAX_THREADS to sane values
* use tabs rather than spaces
Co-authored-by: Kleis Auke Wolthuizen <github@kleisauke.nl>
* spngsave: ensure quantisation occurs last
i.e. after setting the metadata and phyiscal pixel dimensions,
since quantisation will overwrite the image.
* spngsave: avoid over-allocation of temporary buffer
For low-bitdepth write and palette-based output.
* Inverted order of TIFF tag writing when saving multi-page subifd pyramids to ensure that subifd pyramid layers are correctly tagged as reduced-image subfile type and not as pages
* Use if-else clause to avoid tag overwriting
We added a check that loader ->header() and ->load() methods set the
same dhint. This commit removes that check -- it's not possible to make
header and load match in all cases since images can take so many routes,
and can change between the two events.
If we see performance issues around inappropriate cache sizing again,
try to fix just those cases.
* webpsave: switch to sink_disc
* fix some warnings
* distinguish between animated and single image write
* fix some formatting issues
* add missing comments
* fix webpsave_mime
* don't use sink_disc for single image webp write
* fix potential segv/double free
* update changelog
* remove leftover function declaration
* Revert "don't use sink_disc for single image webp write"
This reverts commit a75922ca93ee8987fa5e1ffc52b68e22bd6fb1fb.
* implement review feedback
* adjust to a column size of 80
* get contiguous RGB(A) buffer directly from VipsRegion
* Revert "get contiguous RGB(A) buffer directly from VipsRegion"
This reverts commit 84fc10e52fb223f0208c605dde92d3f821ceee0e.
* accumulate image data directly in frame_bytes buffer
* switch WebP back to Webp naming scheme
* rename page_index to page_number
* revise formatting
Add a check that header and load methods agree on the demand hint, and
make sure all loaders pass.
If they disagree, you can get bad performance in some cases, since the
pipeline can be built from the header dhint.
* move tiff decompress outside lock
Most time in TIFF read is spent in decompression. If we move this
outside the lock, we can get a useful speedup.
This commit adds the machinery to move the lock to before decompress, so
jp2k decompression is now threaded.
Before:
```
$ vips copy wtc.jpg x.tif[tile,compression=jp2k]
$ time vips avg x.tif
117.249845
real 0m15.085s
user 0m16.155s
sys 0m0.109s
```
After:
```
$ time vips avg x.tif
117.249845
real 0m1.207s
user 0m18.384s
sys 0m0.369s
```
* start moving jpg decode outside the lock
* move jpeg decompress outside the lock
seems to work
* add some more tile size checks
double-check jpeg tile size before decode
* fix tiffload demand hinting
We were not setting the hint correctly in header load, and we were not
hinting smalltile for tiled TIFFs.