b183d32b15
use a hash table for tilecache
...
now we can have large numbers of tiles, use a hash table for tile lookup
2012-08-23 16:08:08 +01:00
65f679c7ac
add a line cache
...
add vips_linecache() ... like tilecache, but caches scanlines and sizes
dynamically with request size
2012-08-23 14:51:57 +01:00
fa5677abd7
works!
2012-08-23 13:49:21 +01:00
f5f0dda551
experiment with thread stalling
2012-08-21 16:18:25 +01:00
45e3c0bd39
fix some warnings from jpeg write
...
jpeg write with a non-jpeg source could produce some warnings as it
tried to update exif that wasn't there
2012-08-14 14:31:08 +01:00
88c3f7ab7d
fix compiler warnings
...
some compiler warnings appeared in a production build (assert()s
commented out, -Wall, and -O2)
thanks to Dmitry Roshchin for the report
2012-08-14 12:43:54 +01:00
76fd928d29
more struggles with gtk-doc
2012-08-14 11:20:29 +01:00
368d3c2b8f
test for gtk-doc in bootstrap
2012-08-14 11:00:23 +01:00
bc736e4f5c
better precision for --vips-progress times
...
does 3sf of float time now
2012-08-10 15:19:45 +01:00
1abac4018a
better enum arg error message
...
you now get something like:
$ vips tiffsave k2.jpg x.tif --compression poop
tiffsave: enum 'VipsForeignTiffCompression' has no member 'poop', should be one of: none, jpeg, deflate, packbits, ccittfax4, lzw
2012-08-10 13:54:42 +01:00
463058c149
note and use res unit on jpg load/save
...
previously, jpeg save always used pixels/inch and jpeg load converted to
vips pixels/mm
now on jpg load the image's res unit is recorded in
VIPS_META_RESOLUTION_UNIT, and on jpg save the res unit is set from
VIPS_META_RESOLUTION_UNIT (or defaults to inches).
you can now copy a cm-preferring tiff to a jpg and the unit is preserved
2012-08-07 14:10:19 +01:00
8214d15982
more improvements to dzsave
...
now writes x_files and x.dzi, as per spec (thanks Benjamin)
deprecate tile_width and tile_height, just have tile_size now
2012-08-07 13:29:03 +01:00
0c35f461dc
add "deprecated" flag to vips arguments
...
deprecated arguments still work, but are not shown in help, nor are they
checked for "have-been-set"
you can tag required and optional args as deprecated ... obviously if
you deprecate a required argument you must replace it with a new
argument or scripts will break
2012-08-07 12:52:50 +01:00
a9c7a36670
be more cautious enabling tiff ycbcr
...
we were turning on ycbcr for all jpeg tiffs and relying on the jpeg
compressor to only use it when possible ... now just turn on ycbcr for
8-bit RGB images
2012-08-07 12:17:25 +01:00
4cb63cf1ba
update the C++ binding
...
and the Python one
2012-08-06 22:02:34 +01:00
3276c636f7
add support for VipsInterpolate to the C++ API
...
quick and ugly hack, this will be replaced in vips8
necessary to get affinei and affinei_all in Python
2012-08-06 16:06:17 +01:00
dab22df75b
fixes to deep zoom write
...
- shrink down to a 1x1 pixel tile, even for very long and thin images
- round image size up on shrink
- write a .dzi file with the pyramid params
- default tile size and overlap now matches the openslide writer
thanks to Benjamin Gilbert for pointing out the errors
2012-08-06 15:14:59 +01:00
392b59e261
version bump
2012-08-06 15:14:31 +01:00
3033fea933
add --vips-fatal flag
...
--vips-fatal causes an abort() on the first warning or error, handy for
debugging
2012-07-21 17:01:17 +01:00
133ee84794
fix --without-magick etc.
...
the vips7 compat readers like im_exr2vips() were not guarded with ifdefs
in case their read library was missing
2012-07-20 17:18:49 +01:00
bc13abe3da
update the manual version numbers
2012-07-20 15:52:45 +01:00
7ef115161d
sync
...
finished ruby-vips testing
2012-07-20 15:04:49 +01:00
7df7bb85bb
add --vips-version cmdline arg
...
print version of libvips that a program is running against
2012-07-20 13:34:05 +01:00
a71fe51d2d
support "rs" mode in im_open()
...
vips7 compat mode now supports "rs" mode --- sequential read
ruby-vips uses this to get :sequetial=>true working
2012-07-20 13:07:49 +01:00
41be00d5dd
version bump for new stable
2012-07-20 10:32:20 +01:00
5778dfac1a
remove the version numbers from the .pc files
...
we used to have minor version numbers in the pkg-confiog files, eg:
pkg-config vips-7.28 --cflags
The idea was to make working with multiple installed versions (something
I do a lot at least) easier. But it was also annoying and made upgrading
harder for downstream packages. And the benefits were very marginal.
2012-07-20 10:22:43 +01:00
bb95b54679
oops vips7 png read was broken
2012-07-19 13:59:37 +01:00
c6a714794f
csv load with options was broken
2012-07-18 17:11:10 +01:00
17f6489c0e
Merge remote-tracking branch 'origin/7.28'
...
Conflicts:
ChangeLog
configure.in
libvips/deprecated/im_csv2vips.c
2012-07-18 10:39:29 +01:00
4af1dabb5a
don't cache write operations
...
we want writes to happen every time, in case the file has been deleted
by someone
2012-07-18 10:13:03 +01:00
0b80d436c8
vips7 csv load with options was broken
...
it used the string including the options as the filename
2012-07-17 10:22:01 +01:00
4171f2673d
safer string equality in cache
...
it could segv if operations had strings which had been set to NULL
2012-07-17 10:10:34 +01:00
467a78f32b
save as ppm from vips7 with options was broken
...
it wrote a file with the options still in the name
2012-07-17 10:00:59 +01:00
5363aace39
wopconst was wrong
...
struct offset astray meant that wopconst didn't work
2012-07-17 09:46:16 +01:00
db2c4f572a
png compression range was wrong
...
png compression range was locked to 1-10, should have been 0-9
2012-07-16 15:27:16 +01:00
34602a082a
Merge remote-tracking branch 'origin/7.28'
2012-07-16 14:41:12 +01:00
cd1b215579
vips8 more/moreeq were broken
...
The code to swap the args and use less instead was wrong.
2012-07-16 14:09:03 +01:00
91628efdda
better vips7 option parsing
...
valgrind could complain about the previous one
2012-07-16 13:26:35 +01:00
97424321ac
memory error in vips_sign()
2012-07-16 12:50:28 +01:00
0fc9c286ee
property name mixup
2012-07-16 12:50:09 +01:00
62aa2e3708
oops, misnamed property
2012-07-16 11:28:12 +01:00
88deea9a09
oop typo
...
stray _unlock()
2012-07-15 06:05:34 +01:00
cbe0dcf797
fix another tiny ref leak
2012-07-14 13:05:58 +01:00
73f8263286
sequential can now skip ahead
...
vips_sequential() no longer bans all non-seq access, it just bans
rewinding --- if you ask for something some way ahead, it reads and
throws away pixels
this means VipsExtract and VipsInsert can now be seq: they will read and
throw away any stuff they don't need
2012-07-14 12:13:58 +01:00
5b43d4dd44
get ready for dist
2012-07-13 14:55:26 +01:00
669babb177
sync
2012-07-13 13:51:31 +01:00
071ae38217
better configure summary
2012-07-13 12:56:08 +01:00
3476e39c7e
fix a leak in vips7 support
...
there was a ref cycle being created in file2vips ... fixed
2012-07-13 12:26:51 +01:00
205f13eccb
sync
2012-07-13 09:58:58 +01:00
2c37e77dd8
small tiff save cleanups
2012-07-13 09:30:52 +01:00
John Cupitt