0e0cf1481c
g_free/im_free mixups
2011-09-20 14:52:02 +01:00
4b2c8587b2
vips_malloc() size tracking
...
vips_malloc() now tracks allocation size and can report total mem use.
It seems to trigger quite a few nip2 bugs though, I guess we are
g_free()ing the result in places (or vice versa).
======= Backtrace: =========
/lib/x86_64-linux-gnu/libc.so.6(+0x78a8f)[0x2aae8e011a8f]
/lib/x86_64-linux-gnu/libc.so.6(cfree+0x73)[0x2aae8e0158e3]
/home/john/vips/lib/libvips.so.15(vips_free+0xc2)[0x2aae88f7717e]
/home/john/GIT/nip2/src/nip2[0x4c9ce2]
/home/john/GIT/nip2/src/nip2(path_map_exact+0x63)[0x4ca127]
/home/john/GIT/nip2/src/nip2[0x4b5381]
2011-09-19 16:44:51 +01:00
edb28af67c
more cache fixups
2011-09-19 10:18:48 +01:00
359ce3208e
cleanup
...
file in git which should be blocked, also TODO note
2011-09-12 14:55:22 +01:00
df599f2f55
better vips_init() fallback
...
the vips_init() fallback was not working well ... new, better one gets
called more often
2011-09-12 12:28:01 +01:00
f9b8e0d88f
operation cache working
...
but see notes in file
2011-09-10 14:38:39 +01:00
5d1accfc83
fallback vips_init()
...
call vips_init() for the user if they forget
2011-09-09 17:14:27 +01:00
e7cd1147ce
cache in and working, at a basic level
...
reverted the switch thing as well, annoying
2011-09-09 17:01:29 +01:00
d9883cbd8e
cache fixups
...
fix various problems ... getting closer
2011-09-07 14:10:59 +01:00
bb8f3abf4a
replace ifs with a switch
...
oops, obvious improvement
2011-09-07 13:20:24 +01:00
53b3018f41
add vipsobject cache
...
added the cache ... compiles, but not yet in use
2011-09-06 16:41:37 +01:00
e99f6cc49e
fix up VipsPool
...
vipspool working, test program in ~/try, passes valgrind
2011-09-06 10:37:00 +01:00
22326059ec
add sentinels
2011-09-06 10:36:41 +01:00
774d86234c
oops bad varargs use
2011-09-06 10:35:49 +01:00
b5cadc1815
move C API back to int operations
...
always return results through params
2011-09-05 21:28:35 +01:00
b28cda68a7
TODO update
2011-09-05 18:21:21 +01:00
ad016c53a2
C API supports optional output args
...
you can do stuff like
int x;
vips_min( fred, &min, "x", &x, NULL );
to get the x pos of the minimum
2011-09-05 18:16:32 +01:00
18e2468c26
oops
2011-09-05 14:12:22 +01:00
4ce3346a4f
VipsMin stops early if it can
...
if VipsMin sees a minimum value (eg. 0 for uchar) it'll stop searching
early
2011-09-05 14:10:56 +01:00
6f125a6f31
enable YCbCr compresison for jpeg in tiff write
...
if we're writing RGB jpeg in tiff, enable YCbCR chrominance subsampling
for a 3x improvement in compression
2011-09-05 10:53:17 +01:00
7fe104fee6
support optional output args in the cli
...
the cli interface now supports optional output args ... C API next
2011-09-04 10:10:39 +01:00
bbaef3dad1
sync
2011-09-02 18:21:28 +01:00
958544e5cd
sync
2011-09-02 09:10:33 +01:00
07eb155462
allow optional output args
...
optional output args in C API startup ... need to fix up C API shutdown,
plus CLI, see TODO notes
2011-08-31 11:37:12 +01:00
5b7de5e594
added VipsMin
...
though there's a problem with the cli interface, see TODO
2011-08-30 16:33:25 +01:00
2909cb93a4
add a 'stop' param to generate
...
generate now has a 'stop' param, set this to indicate early successful
termination
this will be used by VipsMin and friends to stop scanning early for
conditions like seen a 0 in a uchar image while searching for min
2011-08-29 14:29:33 +01:00
1a38d60efb
check args more carefully
...
check input and output args have been supplied in different places ...
so ins can be all checked before outputs are made
2011-08-28 13:29:53 +01:00
ac3ce8b010
add VIPS_ARGUMENT_APPEND
...
to give more control over arg ordering, fixes a problem in avg
2011-08-28 12:46:50 +01:00
b26dcb284d
sync
2011-08-27 14:03:52 +01:00
7a392d4789
im_avg() is a class
2011-08-26 10:15:39 +01:00
ae016dd408
turn im_subtract() into a class
2011-08-23 20:56:35 +01:00
7111ff2433
version bump for next cycle
...
s Please enter the commit message for your changes. Lines starting
2011-08-21 18:06:19 +01:00
a6255af981
sync
2011-08-17 09:25:11 +01:00
0b39b4d9c8
better compiling without c++
...
closer to a painless no-c++ build, though not quite there yet
2011-08-16 15:37:50 +01:00
11eb90cdb3
itest for C++ compiler
...
automatically disable C++ support if no compiler found
2011-08-16 13:26:57 +01:00
3383344a31
TODO notes
2011-08-15 18:27:43 +01:00
1cfa33e4fb
better termination handling
...
setting and detecting ->kill is now treated more like a message ... you
can now cancel calculation in nip2 without breaking calculation
2011-08-15 16:24:46 +01:00
1e21963f83
don't use G_VALUE_COLLECT_INIT
2011-08-15 13:59:47 +01:00
f000d93042
fix --without-cfitsio
...
oops, was broken, thanks Mike
2011-08-11 09:35:02 +01:00
b748961743
im_benchmark.c had some debugging left in
...
oops, im_benchmark.c had some work commenetd out, hence the fast run
times
2011-08-10 10:17:36 +01:00
e5effaba94
oop debugging was enabled for disc write
2011-08-09 13:11:03 +01:00
53589638d0
Merge remote-tracking branch 'origin/master'
...
Conflicts:
TODO
2011-08-09 12:47:05 +01:00
b494ce7a52
stuff!
2011-08-09 12:45:42 +01:00
f5397a68af
sync
2011-08-09 11:15:09 +01:00
764ce559f8
better benchmark prog
...
the benchmark program (benchmark/benchmarkn.sh) runs each test three
times and just reports the fastest run
it also detects the number of CPUs you have and automatically loops the
right number of times
additionally, tiles now default to 512x512, so it explicitly sets tiles
back to 64x64 to make comparisons with earlier versions easier
2011-08-03 11:25:31 +01:00
a848adc7b3
add im_concurrency_get() to operation db
...
to help benchmarkn.sh loop appropriately automatically
2011-08-03 11:17:57 +01:00
8790e552b6
fix a race in xyz to lab
...
the table build had a race condition
2011-08-03 10:12:45 +01:00
1bded6682e
todo update
2011-07-29 13:24:52 +01:00
0087b3b655
fix memleak in exif write
2011-07-29 11:41:18 +01:00
122c697349
vips7compat missed old enum names
...
oops, ==0 missing from a strcmp()
2011-07-28 16:52:29 +01:00
John Cupitt