67a5aa7442
make VipsArea leakcheck a run-time option
...
was compile-time
2013-07-16 11:37:01 +01:00
e36d8a843b
sync
2013-07-15 22:01:14 +01:00
6167d4d97c
Fix three minor memleaks
2013-07-15 22:01:00 +01:00
6b3dc259d1
more maplut hacking
2013-07-15 14:57:01 +01:00
440b3a3362
fix up -DVIPS_DEBUG build
2013-07-15 10:49:49 +01:00
0c8ecc4577
oops another const
2013-07-13 13:13:14 +01:00
101c60cf42
add a "const" to get_string
2013-07-13 13:02:06 +01:00
da47510e37
fix some typos
2013-07-11 14:37:53 +01:00
d65b7b269a
add multiscan jpg write?
2013-07-11 09:17:14 +01:00
aadd7c1ee5
vipsthumbnail rewrite
2013-07-10 11:05:45 +01:00
0d00b06731
fix a used-before-set valgrind complaint
2013-07-08 10:29:13 +01:00
ae4922e480
Merge remote-tracking branch 'origin/7.34'
2013-07-06 12:00:40 +01:00
02095763c9
null-terminate exif strings
...
libexif strings are not null-terminated, add a \0 before adding to the
vips image
thanks Mike
2013-07-06 11:51:34 +01:00
c936a86176
starting maplut
2013-07-06 11:49:27 +01:00
1ba188dc83
better file format sniffing
...
and vips_error_freeze/thaw() to stop errors from sniffing
being logged
plus better matrix sniffing
2013-07-04 09:43:57 +01:00
0180a61be4
move im_identity*() to a class
2013-07-03 21:39:28 +01:00
9cc8204ce1
vips_buildlut() works, removing old im_buildlut()
2013-07-03 14:01:46 +01:00
7ee47c2073
vips_buildlut() builds
...
also fix ref doc build
2013-07-03 12:52:39 +01:00
27cec75cbd
Merge remote-tracking branch 'origin/7.34'
...
Conflicts:
ChangeLog
TODO
configure.ac
2013-07-03 11:38:15 +01:00
7bf40144b9
lower priority for matlab load
...
reduces segvs from Mat_Open(), thanks Michael
2013-07-03 11:36:25 +01:00
7312fd4ed4
more array->matrix renaming
2013-07-03 09:22:04 +01:00
7490da07a6
rename image array as matrix
...
we had vips_image_new_array(), this was confusing with VIPS_TYPE_ARRAY,
something completely different ... rename image array as matrix
2013-07-02 22:09:50 +01:00
9278498623
add vips_matrixload(), vips_matrixsave()
...
read and write vips mat format to and from VipsImage
also fix a leak with vips_check_matrix()
2013-07-02 12:08:46 +01:00
e8d9dbe3bb
added array read
2013-07-01 14:49:51 +01:00
8259da2b9c
getting ready to rewrite hist stuff
2013-07-01 13:45:36 +01:00
3338b51871
TODO update
2013-07-01 13:20:31 +01:00
3b6bcef700
strip ICC profile on conversion in xyz2srgb
...
if we use the vips xys->srgb path (ie. not the lcms one) we must remove
any ICC profile left over from import, otherwise we may have srgb pixels
with a non-srgb profile
2013-07-01 09:23:09 +01:00
a818c64f09
Merge remote-tracking branch 'origin/7.34' into 7.34
2013-06-29 13:43:32 +01:00
0b23128ff6
fix two trivial compiler warnings
2013-06-29 13:08:14 +01:00
610f7aa0eb
fix morph on non-uchar images
2013-06-28 14:50:39 +01:00
de796417cb
TODO note
2013-06-26 09:29:04 +01:00
645ff73335
whitespace
2013-06-25 17:49:42 +01:00
824977a0bc
remove the class init lock
...
the previous patch was not sufficient, remove it
2013-06-25 17:46:38 +01:00
bec6d2c8a6
ping all classes in dzsave
...
fixes a race in class build on many-core machines, see
https://github.com/jcupitt/libvips/issues/64
2013-06-25 17:44:57 +01:00
c3ac89df30
missing include
2013-06-25 11:30:18 +01:00
9d5acf79b3
fix ushort save with values >32k
...
we were not setting the fits offset correctly for ushort images, thanks
weaverwb
2013-06-23 04:39:59 +01:00
b58f67aa90
single-thread class init
...
See:
https://github.com/jcupitt/libvips/issues/64
We were seeing races in class init on very-many-core machines. GObject
is supposed to single-thread class init, I think, so this shouldn't
happen. Perhaps class init via g_type_class_ref() is not
single-threaded.
2013-06-21 09:35:04 +01:00
890d7b1205
tiny polish
2013-06-20 10:05:46 +01:00
b20a9f78dd
faster --centre logic for dzsave
...
don't test pixel-by-pixel for background tiles, instead track the rect
of real pixels and test for intersection
this might help performance scaling on many-core CPUs since we are no
longer spinning up a threadgroup for each output tile
thanks Kacey
2013-06-19 14:03:39 +01:00
85eee0528a
fix make dist
2013-06-18 13:49:26 +01:00
d461d082fa
sync
2013-06-18 11:01:43 +01:00
47adc8f08d
flag more str->array errors
...
str->[int] and [double] conversions set the output array to size 0 on
convert fail
2013-06-18 11:00:04 +01:00
2fd2fd58f7
fix up gtk-doc
...
fix various problems:
- out of date IGNORE in doc/reference/Makefile.am
- name clashes between private .h and public
2013-06-17 20:16:20 +01:00
4c0ace34f0
docs cleanups
2013-06-17 14:54:11 +01:00
301d3a1809
argh remove unknown.pot
...
and kill *all* .pot files
2013-06-17 12:05:29 +01:00
dcd2dbb756
vips_pngload_buffer() works
...
fix a bug in vips_jpegload_buffer() too
2013-06-17 09:41:22 +01:00
005984226e
add png load from buffer
...
though untested as yet
2013-06-16 22:51:28 +01:00
1dbff8f5c0
sync
2013-06-16 12:26:27 +01:00
7bdc1e54f2
allow NULL out pointer for vips2buf*()
...
im_vips2bufjpeg() etc. needed a valid output IMAGE*, annoyingly
2013-06-16 10:24:56 +01:00
dad036afbf
make a baseclass for the new create thing
2013-06-14 10:34:00 +01:00
John Cupitt