20ae6b8af7
Prevent non-critical race condition in op cache
...
First thread to add an operation 'wins'
2014-09-19 09:18:58 +01:00
14a8c5ad47
add @profile option to pngsave
...
tiff and jpeg save both have a @profile option, this patch adds one to png as
well, see
https://github.com/jcupitt/libvips/issues/180
2014-09-18 09:38:12 +01:00
e4461ae1f9
fix vipsthumbnail with broken embedded profiles
...
if the embedded profile is broken or incompatible with the specified
output profile, retry with the specified input profile
see
https://github.com/jcupitt/libvips/issues/178
2014-09-12 10:51:49 +01:00
b1f3acf17b
fix vipsthumbnail
...
with input cmyk and output rgb profiles both specified ... it was
importing twice, causing a band mismatch
see
https://github.com/jcupitt/libvips/issues/177
2014-09-12 09:04:58 +01:00
980adb394a
oops, GRAY is an lcms2 feature
2014-09-10 09:47:38 +01:00
01bd7cad04
support GRAY as an input and output ICC space
...
see
https://github.com/jcupitt/libvips/issues/176
2014-09-10 09:37:13 +01:00
8875e23128
fix a setjmp() loop in pngread
...
see
https://github.com/jcupitt/libvips/issues/175
stupid setjmp()
2014-09-10 09:05:13 +01:00
158237254d
fix jfif resunit read
...
see https://github.com/jcupitt/libvips/issues/170
the previous fix just did exif resolution "none", we need jfif "none" as
well
2014-09-09 16:46:16 +01:00
6269c852cc
support exif res unit "none"
...
well why would you want that, but it seems some images use it
2014-09-09 12:18:21 +01:00
d085351f0f
fix mono <-> rgb converter extra band handling
...
see
https://github.com/jcupitt/libvips/issues/172
thanks James
2014-09-09 10:07:35 +01:00
a8becfc9d4
bump version to 7.40.8
2014-09-08 19:47:01 +01:00
0fb1fc974d
Merge pull request #171 from lovell/remove-config-empty-else-7.40
...
Prevent the version of autotools in Redhat/Centos 6 generating empty else statements
2014-09-08 19:35:27 +01:00
a620170362
Prevent empty else block in configure
...
Enables libvips to compile on RHEL/Centos 6
2014-09-08 18:56:51 +01:00
5ab3009ce2
add thread_shutdown() to python
...
see https://github.com/jcupitt/libvips/issues/170
2014-09-07 10:21:01 +01:00
501e900be5
tiny polish
2014-09-02 13:40:37 +01:00
e9ce212b62
make out of order read a fatal error
...
For png and jpg, out of order must always be a fatal error, since you
can't read from a specific y position.
This was a g_assert() before which meant that this test was missing from
production builds. It's now always a fatal error.
2014-09-02 09:37:08 +01:00
70b35f5ec5
update ChangeLog
2014-08-28 16:44:03 +01:00
0eecb1ba06
Merge pull request #167 from lovell/webp-read-buffer-leak-7.40
...
Ensure Read struct is freed when reading WebP image data from a buffer
2014-08-28 16:40:40 +01:00
e09185860d
Merge pull request #168 from lovell/png-save-buffer-unref-area-7.40
...
Correct VipsArea ref count in PNG buffer output
2014-08-28 16:38:29 +01:00
bdf7126e8f
Correct VipsArea ref count in PNG buffer output
2014-08-28 15:54:10 +01:00
e13f6bf6b6
Ensure read_free is called for WebP buffer data
2014-08-28 12:09:10 +01:00
86b729150c
fix memleak in type.c
...
see
https://github.com/jcupitt/libvips/pull/164
2014-08-27 13:46:51 +01:00
30aae507a3
fix memleak in tilecache
2014-08-26 17:07:44 +01:00
a75ddfbd5a
fix matlab load
...
oops, width and height were swapped
2014-08-21 22:01:40 +01:00
04692c29dd
enlarge the buffered tilecache
...
helps vipsthumbnail under load
2014-08-21 15:14:26 +01:00
694d86e5c5
fix vipsthumbnail on very-many-core systems
...
see:
https://github.com/jcupitt/libvips/issues/160
we were using UNBUFFERED, but on very-many-core systems this can get out
of order enough to make vipspng free the read object early
2014-08-21 13:05:29 +01:00
573b94e3ec
more docs argh
2014-08-19 17:57:27 +01:00
af84b3097c
more docs work
2014-08-19 16:57:37 +01:00
03577ad832
fix dzsave date attr on Windows
...
see
https://github.com/jcupitt/libvips/issues/157
thanks John
2014-08-18 21:30:26 +01:00
dfc8ade079
fix up vips_version*() docs
2014-08-18 16:01:26 +01:00
7c8ec68448
more vips_check_init() out of API
...
now private
2014-08-18 15:53:17 +01:00
1abcc1eae4
rename vips_init() as VIPS_INIT()
...
stops a deprecation warning from gtk-doc
vips_init() is now a compat macro
2014-08-18 15:27:10 +01:00
cb2b217434
fix 16-bit PNG save
...
see
https://github.com/jcupitt/libvips/issues/156
2014-08-18 13:54:06 +01:00
0b896119d4
fix similarity rotate + scale
...
thanks Topochicho
see https://github.com/jcupitt/libvips/issues/154
2014-08-13 17:44:29 +01:00
2f72a9925f
more docs work
2014-08-12 13:37:16 +01:00
ce428ffa94
more vipsthumbnail fixes
...
- handle CMYK images
- use unbuffered mode, we have a cache already
- don't try to anti-alias on upscale
2014-08-11 16:48:51 +01:00
ed3c56fcbf
add an anti-alias filter to vipsthumbnail
...
vipsthumbnail blurs slightly between shrink and affine, if the affine
will be doing a large shrink
2014-08-11 10:26:12 +01:00
385b2ea5fa
do argb -> rgba for openslide read associated
2014-08-09 22:21:10 +01:00
296eb8b54e
fix pngload with libpng >=1.6.11
...
thaks Lovell
2014-08-09 17:14:49 +01:00
2e8217338b
docs work
2014-08-09 10:25:34 +01:00
78a50f6ec5
more docs stuff
2014-08-08 17:34:30 +01:00
2b086cfe6e
add signed int bicubic path
...
better rounding for 8- and 16-bit signed int bicubic
2014-08-07 10:13:08 +01:00
2e3baa1bef
more docs stuff
2014-08-06 21:59:13 +01:00
7888526361
fix some compiler warnings
...
various false positives from gcc -O3 flow analysis
2014-08-05 19:23:56 +01:00
46673b60d7
small docs fixups
2014-08-05 10:09:13 +01:00
ebeee822bc
better rounding behaviour for bicubic
...
fixed-point bicubic was not rounding to nearest, introducing some noise
in flat areas
2014-08-05 10:02:11 +01:00
e4996d8dce
don't set shrink too low
2014-08-04 21:25:19 +01:00
f1df91eebb
vipsthumbnail defaults to bicubic + noshapen
...
if bicubic is available
2014-08-04 09:55:18 +01:00
ebec31fe4a
more careful p->t conversion
...
When we convert an image from "p" to "t" in vips_image_wio_input(), zap
start/gen/stop callbacks. This makes vips_region_prepare_to() read from
the "t" rather than trying to gen pixels again.
2014-08-03 17:59:25 +01:00
004c98f296
pngsave interlaced makes a copy of the image
...
interlaced png is hard to write: you have to make seven passes over the
image to make the whole file. It advertises itself as sequential, it
needs to only ake a single pass. Therefore, it must take a copy in
memory of the image to write.
2014-08-03 09:34:46 +01:00
Lovell Fuller
John Cupitt