[Pkg-e-commits] [SCM] Enlightenment DR17 advanced canvas library annotated tag, debian/0.9.9.042-2, created. debian/0.9.9.042-2
Jan Luebbe
jluebbe at debian.org
Sun May 4 20:11:48 UTC 2008
The annotated tag, debian/0.9.9.042-2 has been created
at 2d3d5fe99a7381fe657c54af80870c5dd925ba40 (tag)
tagging cd396efd5aed5776ff80b548316c3a85e6b93270 (commit)
tagged by Jan Luebbe
on Sun May 4 22:00:27 2008 +0200
- Shortlog ------------------------------------------------------------
Debian release 0.9.9.042-2
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.6 (GNU/Linux)
iD8DBQBIHhXhioOL5NhIDy4RAq2HAJwJbEiHBQlTqs17/OGs8Nb5Gi5RlwCg2VLp
RGs+IOuF0RrMkXZWr5feY5c=
=SvAl
-----END PGP SIGNATURE-----
Albin Tonnerre (4):
Actually - don't require pixman, it's only needed for xcb, and we don't build it for now
missing dep on libfontconfig1-dev for libevas-dev
Fix more -dev deps for static linking
partially revert the previous commit. libevas0-dev does depend on libx11-dev and libjpeg62-dev
Jan Luebbe (9):
Import upstream snapshot 0.9.9.040
Merge commit 'upstream-vcs/0.9.9.040'
Import upstream snapshot 0.9.9.041
Merge commit 'upstream-vcs/0.9.9.041'
Import upstream snapshot 0.9.9.042
Merge commit 'upstream-vcs/0.9.9.042'
Import debian dir from old repository
Fix cleaning doc/
Prepare for release
amedeiros (1):
* Fixes segfault in the evas_list_sort algorithm (patch submitted by Antoine Perdaens)
andreas (8):
fixed polygon moving
added some docs with best of my knowledge. Please check and correct if needed.
removed evas-config.in because replaced with pkg-config
- added some API docs
removing some files - sorry added by mistake
sorry don't know why I added these files. Was a big mistake!
please call me stupid! Sorry!
two small typo
andrunko (2):
Make sure image colorspace is properly set.
Updated documentation.
atmosphere (3):
Patch from vac to fix compilation under freebsd by bringing configure.ac
sort your evas lists via a function pointer
build fixes from Spanky courtesy of some gentoo user I imagine, hopefully saving raster some time, if it's wrong, sorry
azundris (5):
* assorted fixes for RPM-building
* spec-file (for RPM)
* potential fix for rpmbuild
* require cqt freetype
* is this sane?
barbieri (52):
Add branch prediction fixes
Use memory pool for rect_node_t
Be polite and flush cache when tilebuf is free()d.
Restrict rectangles to be inside the screen at evas_tiler
Use a single XShmImage and XShmPutImage with clip region.
Clean up and optimize, step 1: unscaled operations.
More image blend and rectangle.
idle_flush
Improve _soft16_scanline_fill_transp_solid(), with bugfix.
Add font/text support for software_16.
Clean up structure, split into files.
Cleanup, remove macros and replace with static inline functions, this also provides soft16_pt_*() functions to paint individual pixels.
Implement image scale.
Clean up structure, split image operations into files.
Name clip structure, make it simple to save and restore clip info.
fix evas software 16 engine build and to add the missing image_data_get method.
Ensure data is loaded when data_get() is called.
Fix compile time CPU detection on ARM
Use dither mask when importing images.
Fix CPP warnings about DM_* macros, do loop unrolling for dither mask processing.
Fix bug introduced with r1.138, removed incorrect offset table.
Support for selectable pointer_mode.
Implement extra operations for image.
Software 16 X11 now does rotation.
software-16-x11 available with pkg-config
Unix is case sensitive.
Do not free NULL graphic context.
Use C89 prototype.
Cosmetic: use correct size for comparison of pointers.
Fix typo that leads to incorrect behavior.
Fix bug of type truncation on 64bits platform.
Cosmetic: use correct size for comparison of pointers.
Use correct type size for conversion between integer and pointer.
Damn, wrong copy&paste fix.
Reorder emission of EVAS_CALLBACK_FREE event.
SDL improvements.
Oops, re->surface, not re->surface->pixels.
Revert SDL changes, I was already commited.
Major rework of blit operations to use pre-multiplied colors.
XSync to avoid flicker
Line, Polygon and minor fixes for software_16.
Simplify image alpha usage.
Revert last patch (Simplify image alpha usage.), it was producing nasty artifacts.
Fix image alpha_set() and fix compiler warning, minor fixes.
Add software_16 cache.
Minor cleanup on software_16, evas_soft16_rectangle.c
Damn typo...
Fix access of unaligned memory.
Missing rename.
Be lighter on 'inline'
Fix possible memory leak with software_16.
Fix bug with out-of-screen paiting in software_16
bdsabian (1):
changed the download links for glu rpms
boris (1):
check that the GLU lib is the correct version (ie has Tess routines)
cobra (2):
Add RPM rules for XPM loader.
Thanks to D. Hageman for the punctuation catch!
codewarrior (1):
fix Bug#: 111
cpk (1):
Wuffdy pupskels.
davemds (2):
Update doxy style
This fix the main page to be displayed. But it really need a content update.
devilhorns (8):
Check that the Textblock_Line exists first before trying to free or else segv
Fix some non-NULL check issues in evas_clip.c . Don't ask how I found these :)
Be a little more paranoid about freeing image data by adding some missing
Fix formatting of a function.
Patch from cedric to fix evas_list_sort issue(s). Keybinding dialog shows
Doxy typo fix.
Fix typo.
More typo/spelling fixes for the README.
dj2 (9):
- fill in the evas_object_textblock_text_get function
- fill in evas_object_textblock_text_del
- rename function for consistency
- get the layout node for the position requested, not the current textblock
- evas_object_textblock_line_get was returning the geometry of the current
- make sure the string isn't blank before appending it
- make sure we have a name before trying to look it up
- this fixes a problem in ewl where you couldn't click to put the cursor on
- don't free the passed in im structure. leave it up to the caller to
doursse (97):
xcb port
some info about the xcb engine
add a TODO for xcb
use of xcb name
remove some warnings
add the modifications of raster, fix some potential bug, clean up. It is still a bit slower than xlib
fix a warning
forget to test if the shm extension exists
missing xcb cflags
evas is now shutdown
fix memory leaks and update to new XCB API
iteration fixes
simplify the detection of xcb
fix warning
fix for the hidden symbols for those who want to test the xcb engine ;)
xcb engine does not depend on cairo
loaders are now modules
fix includes
fix includes
don't load the same module several times
and now the savers...
remove printfs
warning--
bad names
argb target for software xcb engine supported
xrender xcb engine
i've forgotten to remove that :)
missing file
missing x
update to the new xcb api, and a fix
update to the new xcb api
gif loader
tiff loader. saver will come later
tiff saver and more checks
fix a leak in xcb eng sfw (it remains one, but I don't see where it comes from). Update the xrender one. Don't use them yet (but who uses them ??), because it relies on fix of xcb that are in my hd
oups
use Jose patch for soft. xcb engine too
don't use X.h constants, but XCB ones. fix the last leak. fix potential bugs. The last version of xcb is needed
cleaning of the code. Set the title and the class of a window
cleaning of the code. speed up the engine by moving the initialization of shm into the test prog, and by commenting some XCBSync (no problem with that, apparently)
fix a bug and some cleaning
cleaning
don't free a pointer that is not allocated. Clean messages
warnings-- and remove trailing spaces
ignore++
fix xrender xcb engine, add old stuff committed to the Xlib engines and not to the XCB ones, replace deprecated XCBSync call by a call of GetInputFocus, fix some round trips but a complete review of the engines must be done to fix them all, first part of a cleanup, first part of XCB error management
missing reply frees
do not free buf if no available buf
free memory when we quit the tests. Free allocated string in the xcb apps
free the event and don't use deprecated XCBSync function
update to the new api of xcb 1.0. et rid of xcb_aux dependancy
update to the new new api of XCB
update to the new new api of XCB. The xrender engine needs review
automated tools can not be perfect
XCB ICCCM is only needed for the XCB evas tests
replace the predefined STRING atom by its value. It avoids a non core XCB dependancy
update software xcb engine to match xlib engine improvements during startup
pass the default screen to the info, as it is not stored by XCB itself. Speed up a bit evas_software_xcb_x_can_do_shm
including glx.h is not necessary (and should not be included here). Add some missing const
Add some missing const
install evas-xrender-xcb.pc
Fix a leak, a typo and update to the new xcb api
fix trailing spaces and wrong #ifdef
add the DirectDraw engine. Evas is now available on Windows (tm)
add a note for MinGW users
missing declaration function
fix alloca include with mingw, remove trailing spaces
use image cache
add pc files for ddraw and sdl engines
fix the build. Create the shared lib on Windows(tm)
Cedric's fixes
add gif explicitely as loader, to avoid iteration over all the availabl loader modules
remove the CR char. Stupid emacs
Glew OpenGL engine added. Windows can have an opengl engine thgough the Glew library, now. See INSTALL note
fix the case where gl-common is not built if gl_x11 is built and gl_glew is not. gl_common should be check apart those 2 engines, imho
move windows.h include from gl_common to gl_glew
move Xlib includes from gl_common to gl_x11
simplify detection of gl engines
WIN32 is not always defined on Windows platform, while _WIN32 is. Remove trailing spaces and replace c++ comments by C ones
remove 2 consecutive semicolons here and there, ans some trailing spaces
Direct3D engine. I'm not an expert of that api, so there should room for lots of improvements
update todo
use standard syntax when initializing structures instead of specific gcc one
ignore that, cvs
remove c99 structure initialization
resize the directdraw surface when the evas is resized. move directdraw code to its own file
16 bits software directdraw engine. Mainly for win ce users, I guess
Formatting
Add EVAS_MODULES_DIR in the list of paths to search
provide evas-opengl-glew.pc to the open source community
evas-sdl.pc -> evas-software-sdl.pc
indentation
autostuff improvements:
add mail to AC_INIT
fix detection of the glew engine. Add some messages during configure
add --enable-strict option to configure:
don't autodetect Glew
dphase (1):
Changed evas_add_image_from_file() to return NULL if the file could not be
emfox (1):
to include right debian files.
emte (2):
Added the change for BitBake the Standardized Openembedded Build System
Python is too incompetent to parse hyphens
englebass (30):
Use cpp to remove code.
Fix warnings. First compute value, then increment pointer.
Fix warning, first compute value then increment pointer.
Remove unused variables.
Fix warning. First compute value then increment pointer.
Remove unused variables.
Remove unused variables.
Fix warning. First compute value, then increment pointer.
Remove unused variable.
Join the pkconfig frenzy!
Remove unused files.
Remove unused code.
Remove unused variables.
Remove unused variable.
ok is only needed if we build eet font loader.
Write XCB consistently
ignore++
Find chr_x the same way as in the other functions. The main loop is very
Remove unnecessary code.
Remove wrong comma in printf statement.
As cache_key uses evas_stringshare, it should be const.
typo
ignore
Clean up configure.in file
Let the value for AC_ARG_ENABLE reflect the default status (enable/disable).
Use correct value for png libs and cflags
ed isn't enabled by default.
Create macros for converters.
Use $enableval directly.
formatting
gilbertt (1):
*** empty log message ***
hendryx (2):
correcting URL
Starting to add the FAQs, will add others over the next 7 days, and then place the web version online.
horms (13):
added debian stuff maaaaaaaaaaaaaaate
Debian packages may now be built from output of make dist and friends
Pedantry
removed some potential double frees in evas_object_image_*
typo
Build Dependancy: Prefer new libx11-dev over old xlibs-dev
Removed unused *.files and dirs
Reverse bogus change
Put debian files in order
Make sure moc_evas_software_qtopia_main.cpp exists for make dist, even if make clean has been called
Various debian build cleanups
Add debian version
The section of a library's -dev package is generally libdevel
jeoparde (3):
Update README to reflect version change.
...And remember to update README and spec file to reflect the new version.
Patch to update pkg-config from Tilman Sauerbeck <tilman at code-monkey.de>.
kaethorn (12):
split up libevas0 into loaders, savers and engines.
typo.
typo.
Add edb, tiff and gif loaders and savers. Also add module's libtool files to libevas0-dev.
fix conflicting values.
adjust dependency versions and provides.
libevas0-dev depends on libevas0, not on libevas.
add xpm loader to debian packages. clean up the control file.
add package of gl module, completely optional.
add SVG loader to debian packages.
remove evas-config from install files.
enable fb, disable directx.
kiwi (2):
fix some warnings, mostly missing "void" when a function has no parameters
revert this, the index pointer can be negative
kwo (1):
Hushhhh
ljlane (10):
assimilation and.. what was that?
cleanup source names
changelog madness
revert to horm's originals
various stuff
dumb man page
missed another
call autogen.sh instead of configure
something with aclocal, I'll deal with it later
quick hack to remove "/@alocaldir@"
mandrake (7):
hush cvs don't say a word
nice typo. amazing this stuff ever works :)
Hush CVS :)
added two functions I needed in etox
If you don't have GL installed at all, this wouldn't compile.
More updates to make this compile without having Mesa installed.
raster, I know you may not want this here, but apparently this fixes a
mazeone (1):
clean up rastermanglish in the docs :)
mej (21):
Wed Jul 7 16:28:59 2004 Michael Jennings (mej)
Sun Jul 25 17:45:53 2004 Michael Jennings (mej)
Wed Jan 5 00:03:30 2005 Michael Jennings (mej)
Thu Jan 6 10:27:24 2005 Michael Jennings (mej)
Sat Jan 15 01:55:55 2005 Michael Jennings (mej)
Wed Jan 19 17:10:29 2005 Michael Jennings (mej)
Sun Jan 23 22:30:28 2005 Michael Jennings (mej)
Thu Sep 8 17:12:14 2005 Michael Jennings (mej)
Fri Jan 20 00:21:07 2006 Michael Jennings (mej)
Mon Jan 23 05:52:33 2006 Michael Jennings (mej)
Tue Jan 24 14:47:25 2006 Michael Jennings (mej)
Sun Jan 29 12:48:38 2006 Michael Jennings (mej)
Mon Feb 6 21:42:27 2006 Michael Jennings (mej)
Thu Mar 30 01:06:02 2006 Michael Jennings (mej)
Fri May 19 14:52:44 2006 Michael Jennings (mej)
Sat May 27 13:36:20 2006 Michael Jennings (mej)
Sun May 28 15:35:14 2006 Michael Jennings (mej)
Mon Dec 4 15:34:28 2006 Michael Jennings (mej)
Tue Mar 6 13:44:32 2007 Michael Jennings (mej)
Tue Mar 6 13:55:52 2007 Michael Jennings (mej)
Tue Nov 6 21:42:00 2007 Michael Jennings (mej)
monkeyiq (1):
gaurd access to layer->objects->last to check if layer->objects == 0
moom (9):
* The textblock object now gets its color multiplied by the color of its
* Fix hsv-->rgb conversion if h=360.0
* Move the optimizations of Jose for the hsv/rgb conversion functions to Evas
* Add evas_object_smart_parent_get() to get the smart parent of an Evas_Object
* The mouse flags are now correctly calculated when the mouse
* evas_object_clip_unset(): set "cur.have_clipees" to 0 *after* having
* Fix a bug with double-click
* Add interceptors for color_set(), clip_set() and clip_unset()
* [GL-engine] Do software YUV-conversion if GLSL is not supported. I
nash (1):
Make layer get work correctly for smart members.
ncn (6):
Make the Evas docs generation system like the Ecore docs generation system.
Some simplification and creation of some modules.
Some simplification, some new functions docs, some new modules, some new doc bugs.
Some new docs, some cleanup, some more groups.
More documentation, simplification and grouping. Some of these changes are themselves going to require polishing, but the docs are now a little bit more organised.
Just some more docs. A few todos inserted, to be fixed soon.
ningerso (10):
Did some cache miss profiling with a large number of transient text objects
Fix evas_list_promote_list to allow promoting to arbitrary points in the list.
Improve format parsing performance by using preloaded stringshares to do pointer
Reduce additional layout parsing overhead by using the format string without
Fix the build for the FB engine on non-x86 architectures.
Patch from Glen Larsen to fix dlopen failure handling in module loader.
Prevent a segv when setting a image path to NULL.
Initialize the error code before returning.
Fix SDL engine build for case insensitive filesystems.
Fix awk portability by removing the space between the -F flag and its arg.
onefang (11):
If we are checking the DirectFB version, we should make sure directfb_version.h is included.
Nothing to see here folks.
Presenting the Indian Rope Trick. Watch in awe as these young files climb
Typo.
.cvsignore++
.cvsignore++
.cvsignore++.
.cvsignore++
printf added so that we can sort out why it's crashing for Hawkwind.
printf no longer needed.
.cvsignore++
pabs (1):
evas_get_color() attempts to dereference and write to every color
quan74 (9):
fixed arch problem in spec
make dist pkgs build again
added software_generic to spec
fix spec
fixed what I broke in the spec - still needs work to move engine modules to their correct package...
add gif and tiff loaders / savers to spec
changed gif loader BuildRequires so it will still build against libungif
bcond without svg
re-enable svg loader in spec
raster (1113):
Initial revision
'
ignore more
ignore more again
we have......... objects! :) some primitive inheritance too.
i think data structs are almost done.......
some more meat......
more meat
mo mo mo :)
'
tidier
ewooo- gl routines for evas.. been working on them...... :)
blah!
balh1
demo works MUCH better now!
much sexier demo
timing...... :)
blah!
more backedn... yum!
cleaing up code.. loosk much nicer now
framework for imlib2 backend. - ne3ed at least another sample.....
imlib 2 backend works now too :).
back to gl routines
more params......
more routines - working on gl font engine.
1280x1024...
parameters for size
FONTS! evas draws fonts in opengl - accelerated with textures.. and it's
aa fonts in evas1 wooop!
imlib backend does text now too.. god that was easy!
woop! font freeing/caching now in.......... woooooo!
rectangles and lines done both evas and imlbi2 ends
tsted it.... and it works :)
aaaaaaaaaah errrrrrr .... stuff!
gradients work in gl part
whooooppppppeeeeeeeeeeeeeeeeee!
add images........
2 more text geometry rotuines......... um........ done! :)
wooooo!
fix trails
cleaner use func pointers for rendering backend funcs - too much copy & paste
a bit cleaner now too :)
add data :)
check for gl & glu now too - better checks
eas now handles fill params for image. tile and offset of image to be filled
dont need bits code.. that will go in another lib - higher level than a canvas
evas detects gl and build in gl backend optionally if it finds it. evas will
text objects work! dude! like totally! :)
lines and rectangles work. woop! :)
gradients added and work
default fotn cache and image cache for imlib subsystem are bigger
oops bugs in line backend for imlb - fixed :)
clean clean
endianess chekc for texture gneration
test uses evas inlcudes in tarball not system
better gl lib/header checkign in configure
api filled out much better now........ infact... i'm almost done :)
add that
much better - fix that function name
get geometyr right when changing text
event system works - callbakcs happen.. WOOHOO :)
big click me to drag text object
handle events better on other property chnages
and prinfs out the fps :)
add to the README
fix line draw bug - add event pass through flag for objects
evas-config scritp added
fux fux fux
missed those libs
add some more query calls.
get rid of replace
missing border attribute setting on render.
slight stakcing chnage redraw problem - didnt queue a redraw if stacking
double free of callbakcs! no no no - don't do that!
silly - move that free
dont need that file
add font ascent/descent checking
ok- work again.. was just testing..... :)
routiens for mroe text info.....
add the api routines to query the text objects
we have a new rendering target for evas - an imlib image buffer.
dissalow resizing for line and text objects - doesnt make sense -
dont call visual and colormap inits
same here
only init once
allow color allocation policy to be set in evas
apply some selly's "no more leaks"
fix minor problem there.........
comment added to evas_free - compiler bug? eh? really odd
ok - that was screwy.. :) propt type problem - fixed :)
x11 engine code added - only partially works
ok - thats a litle better perfromance
rectangles now work in the x11 engine
fix that mask leak
lines now work
much better! dithe rpatterns match for eect updates - fix leak
blah - test that out :) x11 evas does NOT do well in evas test... but
mo mo
add code to double check object type before doing anything to objects
make test programs help output be useful
fix mimpmap bug for wierd scalings.....
new test program started work on (old test is evas_test_old)
dont need those pics
update functions in oepng-less empty prototypes
more on the new evas test. add some more api calls.
privatise data structs for evas... much cleaner. all structs are now internal
update README
some basic start on documentation....what's happening to me? all this
i forgot what i did - but it's good whatever it is....
start of polygon drawing support...
attempts at making polys work all over - gl engine has problems though...
polygon demo
fix that
add images
working on evas test program
do total average fps
test program handles color alpha blended mouse pointer a little better now :)
final lump - added images for image demo
flying action........ :)
back to normal
slight pointer inconsistancies fixed
some new api calls :)
more stuff........... :)
fix gl engine bug with text rendering... hmmm odd... :)
oooooooh text! :)
nicer text flow
more cleanings
lets display fps...... :)
add ungrab call for when butotn is down on object but u want to have it
intro looks even bettre now with images fading in and out....... :)
try indiect if direct dont work
add cheese
mouldy cheese
cheese went rancid
typo
region obscuring code. this should provide for apps using evas to indicate
evas_test client is smart enough to figoure out what windows obscure it and
text comes in a little nicer now
make sure we bind AND then set th minand mag filters
some............ MEAT! :)
image renderer works withotu needing display info
rects and lines display :)
added save file flag for old evas test and fixe dimage buffer rendering to work
commit pabs's fix for -ansi -pedantic compiling
add external dir specified for support libs to LDFLAGS at configure time
added event paranoia for when deleteing objects - makes damn sure the grabbed
more event paranoia. should all be fixed now
fix mouse enter bug for first enter.
commit version up for README
stubs for Xfree86 Render extension support. STUBS only now. will get filled in
test
add some code ro render routines to hanbdle render extension... not much yet
xrender configure detect
and only include render header if we have render....
make sure that make distcheck works
minor documentation fixes
just testing new commit message
test commit
update the documentation to include a good explanation of how rendering works
update a bit of the readme
raster: special case optimization for retcangle objects... since they are a
evas_test now make installs :)
slightly more correct :)
add files so they get into extra_dist :)
make distcheck pass.......... :)
fix a minor event bug with retaining the wrong coordinates fomr the last
fix "special case" for rectangles - property chnage parameter wasnt being set
fix bug with gradient texture binding :)
add ungrab call for ungrabbing the implicit grab when buttons go down....
mouse out event - ay object mouse is in.... leave event... generate.
return values in default cases where applicable
enable the font bug again... debugging dri.. :)
slightly cleaner code... :)
increase font cache and image cache.. this shoudl help a bit
woooop fix fix fix.. note to self... font caching in the gl engine REALLY
doesn't pass make distcheck!
dont return null fi images doesnt exist.. obejct creation shoudl ALWAYs
add new render call that returns the updates that got rendered :)
and return valid info when theres nothngi to do... :)
I wonder what i fixed.... :)
evas_get_layer();
* rewrote GL engine for evas - vast speed improvements and cleanliness
partial buffer swap code gets coords right - but GOD is it slow... i think
x11 engine now does fonts :)
tst
tusting
This finishes up the complete rewrite of evas's backend engines to be cleaner,
up to 0.0.3
test
test
a test.....
tst....
tststststts
fix that event problem... and errr.. some gfx optimizations :)
errr - beat me dead with a large possum - but somehow these 2 files got
* fix requires for imlib2 to require 1.0.1
minor fixup if last char in string has 0 boundingbox width :)
well if u have to have it in the makefile- have the files!
evas.m4
ooooops - fix gl engine to handle window resizes..... :)
more optimizations in render function
load_error now stored and retrievable for image objects :)
oops - minor logic hiccup - fixed.. also imlib2 fix for its load error
and handle that other case...
get rid of those printfs.
damnit... the more i use evas the more i optimize it... ok - more
oops- over optimized so that it didnt actually render some changes :) fixed
oops - again.. over-optimize
0.0.4
err forget it...
forget that forray into optimizing that...
cache image scalings... :)
oops missing that.. fix
ooooooooooooops... fixed :)
let me just turn off scaled caching for a bit.. i need to think this over a
dont allow viewport or render size to be < 1x1
commit :)
1. optimize evas lists a bit by keeping a "last" pointer
god daaaaamn! fix fix fix fix :)
slight "over optimization" :) fix fix
oops coord transforms form world to screen and back were um.. 1 letter off :)
image cache for eva sis preciselyu that.. cache. image and font data are
and invalidate held data if u change image file or font (font/size)
and set handle to NULL when we're done freeing it....
oops a printf.. remove
fix that minor oversight for scaled ouput evases...
add call to allow program to query if image object has alpha...
and that woudl fix a segv problem.. wheee..
oops leak.. fix fix fix :)
optimize... optimize...
debug debug debug.. fix :)
up to 0.5.0 versaion for evas.
fix soem routines that i found incosistent whilst writitng the docs... :)
new docs for evas....
oops - didnt translate viewport for obj's other than images
update docs...
oops double scaling for viewport - fix fix fix :)
oops printf.. remove
updated evas docs
err...
try this
-lm there too... damn. i wish systems woudl use shared lib dependancies. all
now just be cleaning things a bit i got rid of that rogue -lImlib2. i disabled
now that was just silly.. evas shoudl have used the visual and colormap u
optimize! software enigne image objects will keep a pre-scaled copy IF:
ewwwww.. i forgot to free the font string....
ewwwwps.. set pointer for list to NULL after its freed! :)
oooops.. i forgot one lump of logic for adding new object if currently the
and no more printf's
oooops... doint turn off blending just becuase theimage has no alpah
well it woudl appear that only rendering parts of the dipslay using clips
allow for doing bounding box only.. hmm damn you gl.. why suck when all i
offline working... woo.. woo..
inconsistent naming fo func... fix.. and fix docs.
hrrm. a tad more efficient :)
ewwps! :)
save some ram by backing structs :)
oops :) thanks :)
oooh fix fix :)
copy over ft1 support too... :)
you can address objects by name now all image objects by default get a name
forgot to ever do this.. mayaswell reload changed image objects off disk...
oops - bugie :)
oops - fix :)
nuke
in goes new code
code move
change proposed api...
docs on all the evas_list* calls and a few more for completeness
code code...
the big internal function call renaming happened... and it was good.
more documentation.
update authors.
fix till :)
docs... and memory error stuff... auditing memory allocs and making ti very
work with latest dfb in cvs
spotted some old debugging code... and fixed up dfb engine to handle sruface
sellies no-more-segv's
more documentation.
some comments...
rico-suave.
update todo...
todo list has version numbers to do features for.
add build instructions to docs...
update...
informers & interceptors... meant to "inform" callbacks of when an object
informers now are just event callbacks. makes for a simpler api. interceptors
keyboard support... key grabs dont work.. and i'm not happy with it as an
oops.. segv in dfb driver.. fix fix
code code.. fix
blah blah
oops.. return POSTION, not glyph value!
hmm. forgot to translate text coords form queries.. and some glyph vs char
and... fix the smooth scaler.. i had a 32bit overflow there for when images
fix dfb window test proggy.. flip properly :)
maybe debians include system might be happier now...
ok.. a dfb bug.. i have to safeguard...
bah bah bah.... 8-P~
commit...
where did this go?
todo madness
key grabs are done... ok not as efficiently as i'd liek so having more than
evas_test will now tell u fps...
ok. good stats now
PROPER benchmarking now...
ok - dont disable printf output... for debugging
apparently i updated something...
oh my wordy lordy me.. too accurate for its own good soi've had to fidge to
notes in README
aaaaaaaaah freetype 2.1.3 bugs TOO! bloody hell! ft gets BUGGIER as it goes!
resolve symlinks...
and dont leak
const-ify lots of things...
logic problem on evas_free(). shoudl be fixed now... :)
dont need fmemopen suppor tin evas now... just sue eet directly and dont
ok ok.. callabcks walking lists.. delete flags... you get the idea. (i've
make this baby an int
and lets make that safe for smart obbies
after some testing.. this code has been found wanting...
then there weere grabs
code
symbol fixup :) (conflict with imlib2)
fear the changes... api breakage... everyone.. errr... fix yer proggies! :)
up to pre5
freeze/thaw event yhandling... and dont respond to a changed focus after
hmmm breakage.. not good.
ewps!
oops... bug bug bug... FIX :)
oops... fix fix
update spec
we build nice rpm's
i've been leaning mem usage down... save 30% on all evas_objects now! :)
softare_x11 engine can now rotate on the fly :) ecore_evas relies on this btw.
major oopsie. :)
now that documentation is looking nicer
add img
update docs
alignment...
Evas doc update... at least all api calls are now listed
bauil out of a double-init
avoid a cmp/bra
oops cut one line too many
tab...
missing ifdef
memory buffer engine... not doing everything. api may change... working on it.
and configre check right...
only tar up docs ONCE!
buffer engine seems to be wroking for an rgb buffer with color keying...
special case fill if fill = object size to avoif roundoff miscalcs
fix
allow for a callback rtoo to provid e adest buffer when needed
packageconfig support
accept NULLS for font queries
viewport bug fixed
fonts.dir fonts.alias support... so u can point evas at x's font
bump up to pre_6
symbol clash! oops!
evas now uses
change of types
wheel patches
shared lib deps only now
minus 1 printf
kerning fix?
test...
TAKE THAT BITCH!
mo' testiclies BIOTCH!
multiple bitch-ass ho's
nicer formatting
testypants
aaaah DONT use inlined bitmaps.. we want it ALL aa'd
new calls to help string processing for programs - these walk up and down 1
fix corner case of starting at nul terminator byte at end and getting prev
pre7
more pre7
fixing minor arithmeic stuff in text queries. all happy now.
oops leakage on ferring an evas! fix.
get rid of minor flota->in calculation artifact i found!
some optimizations seem to have screwed up rendering in corner cases. fixed
fix fix fix :)
viewporties?
try this?
make deleting of grabs within a key callback "safe"
break!
ipaq build stuff :)
cross-compile properly
optimize font searching... fixed that fixme.
optimize text object state change.. much better!
a little nicer
bleh!
valgrindy goodness
also if text itself changes...
handy conf script
update spec
make packages
packages
ooooops. evas bug1 fix fix fix
package pants
oops overzealous optimizing made me miss a case... naughty naughty.. very
oops. on evas_free nested smart objects might just not clean up right.. fix
pre11
GL engine goes in... its not 100% complete. fonts are sub-optimal. i quickly
debugging, printfs, turn text off for now, etc.
and lets not keep nvidia extensions in code...
query extensions now... properly... :)
pthread too
split gl common routines up into seperate files. still have BIG problems with
open gl is fulyl functional now - it coudl defnitely do with optimizations
1. mmx2 pixel copy and cleanup of pixel copy routines
improve optimal nature of code.. get some more speed....
clean
dotn segv/
well cpu detection was fucked up.. fixed now! :)
compielr prefes this
get rid of useless shm check.... and other things (cant remember)
do autoconf options properly
move co-ordinates to be their own type. this is a good thing, because ti
oops.
dont segv if font wasnt loaded and we query
8bpp rendering sucked. now it doesnt suck as much.
and 666 dithering was arse. fixed.
yuv pixel import api (well arbitary pixel format import api) for doing media
and add yuv conversion code
and set alpha bits
more comments
propagate events from child objects to msart object parents
update docs to include suggestions for yuv
shutting down font systems eh? :) that's cool - but lets make it nestable for
till's shutdown patch - this is good (tm). :)
no no no!
added in loading froms from memory buffers at the engine level, and now an
uninitialized...
ensure these list heads are 0'd out
and dont leave stale fonthandles around...
missed a size_use() call that would have screwed up some font queries... and
2 problems fixed.
and fix the sizing bug!!!!! all happy now!
fix acent/descent
and fix that
and now... AUTO configure.. no need to --enable anymore... evas's configure
dont need this anymore
freeze/thaw debug!
sorry formie - we want these in! you have to find the extra thaw!
oops - got mmx and sse mixed. fix.
split up nasty smooth scaler code to separate parts - they are #included
dfb duplicates common code.. why? why not just call off to the common code.
gl engine bugzie! fix!
oops - delete_me flag!
oops --disable stuff wouldnt work! :)
patchies in
update docs a bit
if outbuf != multiple of 16... clip
cross-compile cleanly damnit!
oops typo
some more docs...
@'s
gendoc
fix multiple gl canvas issues
ah! that was silly. fix!
check deleted flag.. need to do this in more places tho....
fix stack queries! :)
a good 20% speedup on my p4 and minor speedup on this p3. no difference on my
oops - this DOESNT actually work - i thought as much! i knew i had my
just in case this gets through...
try bigger buffer for prefetch test?
jose's evas patches are in.
booo - trying to debug
distable... :)
oooh siglongjmp!!!!
SOME of the changes needed to support evoak... needs new eet...
font imaging stuff....
as an excercise in forcing everyone to write code using the correct
avoid type overflow...
configure portable goodness
hmm math just a little off
nuke #warning, and return no more!
lets limit the starting defaults
extra paranoid checks for text ptr
and check there
lean down memory usage per process - now it doesnt use up 64kb it doesnt need
-lm for libpng
oops - missed init func entry points. fixed.
-lm added
vera... no more non free fonties
add!
cross-compile fix
hot damn! yuv -> rgb was being done TWICE every time! once with mmx/sse and
use mmx2 routines if we can - they are faster! (almost 3 times)
fix palette setting..
propogate pass_events to sub-objects of smart objects...
force fullscreen edraws for gl - this holw buffer swap thing just doesnt work
better autofools
in!
oops
and here are the stubs of a cairo engine
mo files
ignores
fix to account for new engine
valgrind problems... valgrind told me so! :)
reduce stat calls massively...
oops. printf left in.
palette + transp fix
oops copy & paste error
oops- for repat events this screws up grabbing..
fix up grabs a bit more :)
bernhard's patch to handle new style device names as well as old.
mouse out/int do good things now
no printfs
hiq patch
brett's patch for resize and doc typos :)
null patch...
fix mouse in/out grab issues
oops printf's
clip bug fix :)
dont wipe filename if getting pixels for reading :)
load - check! :)
and invalidate that image...
and dont reload if the image is generated pixels...
this has always been there - it measn xft, x and evas dont agree on the size
oops - NOw they agree. evas is fixed to 75 dpi - thats how it thinks now.
only pass certain callbacks to smart parents...
and dont propagate those...
if we dont set LDFLAGs... why not just use the defaults? :)
dont need to set LDFLAGS
add this sucker.
distcheck
buildie cleanies
ldflags
openembedded build files... this makes life so easy to build efl for
optimization flags
aaaaaaaaaah. rectangulkar textures... fix fix... :)
fix mouse grab accounting on delete and hide
hidden symbol support
fixies...
extend key and mouse events
change order...
1. move font dir to its own
and name the fn better
some stubs.. thinking about how to do this...
fix oops
buffer engine fixes
oooh... i found a nasty mouse grab/count/accoutning/hold etc. issue with
useless data files removed... making room for improvements in the engine.
well it was an interesting experiment. but the blender code is already so
span lists were a failure. sloweer than the current brute force method - and
been working textblock...
the textblock actually works.
1. found leak in event code - fix.
fixing up textblock. does horiozntal and vertical (line) alignment
some notes as to what to do next
some extra comments for me to work on...
massive speedup in textblock... maaaaaaaaaaasive.
and in one foul swoop i merged font loading code betwene text block and text
more todo items for textblock
some more comments...
BAD tilman! what if the list is NULL! :)
and fix that too
wherd wap!
and get ourselves another 10% speedup with textblock... :) (and in fact pave
avoid infinite loop more aggressively?
dont xloop...
werd wrap is better. it remvoes the whitepsace OVER the wrap...
update TODO for TB.
optimise... THIS! BIATCH! :)
forgot a line from my testing stuff for perfromance
i worked on a regionbuf set of code (exact rectangles). i THINK it has some
centralise tile size to a #define.. move to 8... :)
underline, double underline, strikethrough and backging color support.
see diff. :)
get font sizes right even if text object is NULL string..
more work on textblocks! :)
fix format size calc - missed last line.
fix some more stuff
next.prev char works.. a lot works. now i have to test text insertion at any
fix shaped output (rgba dest rendering)
a possible smart bug?.... and some textblock notes
a note for moi
fix segv in textblock.
minus some printfs'
working more on it...
complete more of textblock
more work on textblock. the api as it stands is now complete enough for all
and dont segv. :)
remove api that isnt functioning and wont for a while.
working on styles - found bugs! :(
more things taken off the list of things to do. styles seem to work.. :)
tabs work...
fix some native size stuff...
fix newline with a font (size) change. add tabstop formatting... start on
fix some valgrind nasties - margins work...
change versions. some went DOWN - because they arent releases and i'm trying
1. e17's init icons get put into an e_box for arrangement nad the init splash
detect eet/edb properly
bernhard's font scale patch..
and declare that variable before it is used...
off by 1 native size fixed
eet/edb cflags
auto-package for evas and cleanups
add this
callback paranoia
fix packaging
remove TODO
libevas0... use matching versions
fix! textblokc word wrap and spec.in
include ctype.h
werd to my homies... this had to happen sooner or later.... evas supports
use engine add's
oops - size properly. fix!
buidl fix
asparagus time for some of the core efl bits...
let the configre/options determine mmx buildability
eet image load logic was reversed! it worked anwyay - just nto as
ooh issues! thanks jose!
and if the image size isnt set yet...
asparagus!
cleaner disabling of valgrind
fix flags and libs setup for options
fluffybrett!
support unfilled borders with images (middle doesnt fill)
dont realod ye demons of evil slowness! FIX FIX! cache you biotch! cache!
work on valgrind issue... this should fix dropshadow - evas didnt allow us to
optimisipoos. (e17's menus should be more responsive now)
fix detect for k6 cpu's
detect round 2
over-optimise. this broke things
david stevenson's tidyup
build with amd64 asm (needs to be manually enabled though right now)
dont always use backign buffer in fbs < 24bpp
patch for fb engine from
fix list bug...
make e17 pass make distcheck....
delete a smart child from its parent if it is still in there on cleanup :)
fix font geometry issues...
allow bitmap rendering...
asparagus!
full asparagus
clean up software x11 engine a little. buffer converted pixels and copy to
need 2.1.7 or higher...
whitespace
whitespace
warnigns cleanups + fb 32bpp transparency support for evas from celement
some debugging commented out. seems evas's caches work like a charm
make hash fn a bit better
maybe gcc4 is happy now?
asparagus...
aspara!
tackpatch
tack patch
it's been a while, so it was time for some ASPARAGUS on our plates
tb fix!
1. id3 album cover loader patches
one way or the other i'm forced to add timestamps to events - i cant avoid it
natah's clip patch - seems to be all genki :)
shh!
apps/e/enlightenment.spec CVS: apps/entice/configure.in
engage seems to be generating clip weirdness - testing to see if the clip
2nd asapargus for the weekend :)
fix pointer staying in bug.
i have started work on new textblock api/code - it is (for now) known as
doing more work on tb
ned textblock code finally formats and displays.. SOMETHING. this code now
word wrap works (mostly)
wrap correctly
printf--
get rid of most whitespace. still need to handle end conditions
working on tb2
printf--
textblock work
aaargh! dealing with lots of little niggly things - and i havent' even
tb2 is now mostly behaving with formatting. excellent.
margins... werdibubble
and double tb2's layout speed...
valign and parse styles, underline etc. no rendering ... yet
handle tabs.. (jump to next tabstop)
and well.. now tb2 handles all the most important formattign tb1 did. that's
fix valgrind errors
ok - fix up stikethrough over linewraps
remove debug rect
some more work on tb2
more tb2!
you can query certain format nodes - like \n and \t... :)
oh STUPId me - i forgot to add current y co-ord for ptr offset! bah!
check for glue! :)
+ )
fix parse error when there are no tags... :)
fix bit order for ppc - cleanly.
temporary smart stacking workaround until we fix it properly with internal
some extra checks and leak fix
tb2 work
fix tab line wrapping and make format append/prepend work properly with
and fix format prepend
remove debugging rect
fix some leaks, bugs and add api calls that work! :)
more tb2 work!
tb2 work
ok ok - they look ugly, but handle bitmap fonts! :(
fix some whitespace wrapping mojo
tb2 work! :)
more tb2 work! :)
text range get works. u can gety it INLCUDING \n's and \t's (simple
hmm thats better - cursor to STARt when settign markyp
handle unfound fonts
accept escapes (\\, \", \', \ , etc.)
dont need quote games there. we have escapes
printf--
um cleanups.
more escape handling
actualyl don't need that. it's already de-escaped
unneeded var
handle null string
fix tb2 move redraws
fix spaces at start of nodes and lines
a bit of asparagus action for shits & giggles
oops - fix freeze stuff.
fix fallback handling! :)
jose patch! :)
dfb patch
fix some stuff for altivec on linux latest gcc 4.0.2...
disable dfb cvs updates for now
fix grab handling
asparagus!
an xrender engine. fully complete - but it doesnt support shaped window masks
add makefile.am
fix gl engine - i broke it a bit
after some quiet on the western front - asparagus.
in theory support dest alpha
update mask support...
fix target mask support
fix null data copy
fix xrender engnie to accept user-provided pixel data
fix mem issue
fix gradients on scale
fix pixel uploads form emotion
oops - accidentally added this. thanks dj2! fixed.
be more paranoid about poly point array and lines.
was playing with ansiotropic filtering - with mipmap stuff it looks nicer.
endianess fix
asparagus. and make distcheck passes again.
turn off dither and component alpha
and no dither and ca there too
support clipouts in xrender engine
playing with gl engine :)
remove diplucate pixemaps!
handle shm addr return right
fix segv in shared data chunks
use best filter - :)
asparagus - again. lots fo leak fixes and other fixes have been happening, so
dont allow non-defaulted formats
the first wave of evas smart object modes. commit hell #1. :) other things
fix restack draw bug
no that made it worse.. evas renders everything. fix. must find it.
fix renders now
burn witch... BURN!
empty string SHOULD work now if the node is non exiustant to start with
no more old textblock - gone. completely. tb2 renamed (as threatened) to
linux ppc altivec patches - and some tiler speedusp while i was hunting a non-bug
add calls to tell evas to propagate (or not) events on an object to its
fix hash calls, fix object leak. :)
fix excess overdraw bug
1 more over-render - same concept.
and a few less allocs to improve evas_hash
dont lose context pointer
fix leakies
add a call
i think this will fix the layer thing nathan found
1. diable a bunch of engines from auto-detecting stuff. you have to manually
add the file to cvs
asparagus time! :)
ok - freetype does WEIRD shit setting grays to bizarre values so now i have
i'm playign with memory pools for evas lists. it seems to improve things.
--c++
disabled for now - need smroe tuning/testing/optimising
enable pool - nicer on mem
turn of mempool for now
move pow lut table to read only shared memory
new - much cleaner/simpler mempool
reduce a few more alloc counts by merging
typo fix
clip colors apply to style colors now too
asparagus!
dont allocate 0 buf
use stringshare. saves a few hundred allocs... if we start doing lots of text
use stringshare too
less overhead per string for stringshare (overhead right now is 2 pointers, 1
disable stringshare... something fishy.
asparagus. there were issues. thsi syncs versions now too.
stringshare works! :)
cvs ignore++
disabled debugging.
some complaining if we do something bad
remove 1 pointer overhead for stringshare (4 bytes or 8) and optimsie free
oops fix stringshare conversion bug.
oops - textblock badness with alloca
joses's gradient work - gradient look nice. one problem jose.. USE BRACKETS!
only check for c++ compiler if buildign qtopia engine
gl patch - build.
use the right symbol
use djb2 hash for evas_hash too
bloody valgrind! argh! :(
1. good segv catch by whereami.
better style (more files to look at though)
valgrind complaint! fix!
i have to back out all of jose's blend changes - musch faster than
well i don't know... BUT... amd64 here (x86_64) DOES handle mmx, mmx2 and see
fix segv with bad input data
actually make symbol hiding work!
move headers
vis -> off
big fat asparagus!
buffers man.. buffers!
buffers dude... buffers!
correct fmt strings
patch from al catching nasty oops
initial import fo loadable engine modules for evas - from jorge zappie :)
anmd fix dlopen link detection , actually check for dladdr as it's a gnu
remove printfs
oops - a little too much copy & paste :)
oops - forgot about engine headers. moved headers. fixed. install again
temporary fix.
ok - now support a first try guess based on extension - if that fails, or the
removes some unused struct members (should remove more though) and fixes up
most oops - dont strlen that buf!
bad buf strlen--
fix buf free and some extras...
free module paths on shutdown
empty buf first
fix dfb engine - correctness changes for hash
no! dont put that in cvs!
remove cvs ignores!
clean up new modular build to link to the right things, in the right way,
clean up new modular build to link to the right things, in the right way,
fix file lists.
fix spec file to at least indicate what i mean for the "best way" to package.
fix stray ~
add cpp
actually use the sse routines!
oops - missed fixup of end ptr. :)
1. buffer test really wqasnt being very fair - time to be fair.
sse detect b0rken. fix.
smaller buf for sse test (was too big anyway) and mmx2 is prettry much sse :)
fix egetnv of $HOME null return
text objects never drew with offsets to account for style padding - done now :)
disable double click flags if u clock on something else in between
evas lets you choose font hinting - in theory. :)
oops - yeah. watch the ptr.
make hinting an engine api thnig - engine apis need to be mroe easily updated
rehinting works on the fly now
rehinting works on the fly now
1. new engine inheritance scheme - its recursive if u want it to be, and so
software engines all now totally inherit from software_generic and
oops - nuke a \
oops - missing dirs
keep the function call naming consistent with other engines of the same type.
and build
fix symbols
oops - soemoen changed the other symbols around while i changed the first set
umm... handle rgba dest properly...
check ptrs - no segv
add a dst alpha flag
ARGB targets for sw engine supported
pass distcheck...
asparagus!
nooo. debian/changelog is meant to be in MAINTAINERCLEANFILES - because make
tmp size was not fetched
asparagus!!!
cvs is back up.. time for some asparagus!
zero out alpha only if we have it - and add an initial memset to 0.
fix deps
cvlear ALL dst buffers for buffer engine
support all advertised formats
we dont need those filea at all.
and put files in Makefile.am!!!
bleh--
ok- actually handle dest buf if there is no back buf
gasp! packages actually build!
add fc flags
fix maintainer
fontconfig mem reduction patch (only load what is needed)
fix resize of shaped xrender canvas
fix color multiply whiel blend (ergh! need temporary image! yuk!)
fix edge interpolation for images with xrender and a segv for font rendering
no - the .a and .la's for moduels ar enever ever used - even for devel.
and .so is in dev
and other .so's are here
we dont need evas0-bin anymore - and the moduels may create subdirs of any
remove file from makefile - removed
fix over/under grab problems
fix evas spec to build rpm packages with detailed moduels & requirements so
fix evas spec deps
fix packaging - edb not built by default (frankly- nothing uses it!)
cosmetic - really no diff
move engine headers into engine dirs - and removes lots of useless old headers.
jose's software rendering work - slight improvements (about 5-10%). i had to
gl engine fixes patch from jose, and fix non mmx/sse system build
fix "fades" of non-alpha images.
turn off vsync
fixed mmx.h from gfx_routines
fix mmx more
make xrender transforms use 16.16 matrix values (xrender seems infintely mroe
weird - identity matrix with 16.16 fixed point DOESNt work for bitmap mask
illogict patch for more escapes :)
usless call- comment ou
really simple fix for aa line draws.
fix xrender lines :)
fix polygons too. always over op.
fix segv's in simple line draws
'remove fps stats
fixup for over-grab-count
1. fix ecore to build again (xcomp - u need to cvs add and commit your 2 new
fix sizing problem when u set style after setting text/font
patches that i said were in - commit. (see my reply emails)
1. autofoot patches.
fix blah-config includes
sort patch from cedric
revert list sort patch - it's losing list members.
another good time for good asparagus :) snapshot.
remove openembedde pkg info - old and dead
asparagus time...
better configure checks mean that evas will build on rhas 3
way too many freetype headers are wrong - dont rely on them. :(
asparagus time!
edje prefix fixes - oops - was broken
png check fix
fix png check - succeed/fail clauses swapped wrongly.
make other core efl things do the same package thing
libpng checks using pkg-config :)
aspara!!!
release the asparagus... another snapshot! :)
add 2 list calls that well - i need for efficiency :)
and that was just SILLEH! - fix. fast fix.
add abort so u can trap etk badness
add lib deps back in - seemingly i found a modern gcc/ld that doesnt seem to
jose's xpm loader.
fix some warnings!
being an anally retentive bastard.. i checked if e worked in 8bpp under vnc.
remove printf
jose's svg loader - in. needs the ability to set dpi etc. etc. - loader api
fix subdir!
evas svg loader .spec file additions
change cwd to the dir with the svg - if the svg has relative paths to bitmap
memset to 0 for initial memory
added the ability to set load scaledown, dpi or load size - the svg loader
and actually use the load size
guess svgz faster too
fix caching to account for load options
jpeg loader can use load opts to try pre-scale on load now... :) ie epeg
printf-- - safety valve++
include evas private header
gl engine build
use already loaded var :)
check for fedora cairo svg thingy
a few patches from emails. :) emails will be coming.. when i flush my mush.
add system icons dir for e, patches for evas.
svg fallback fixed
#ifdef the 1 bit transform stuff.
and i implemented module UNLOADING. given enough cycles of non-use of a
tune the unloads to be conservative but effective.
avoid loading all loaders if file doesn't exist
cleanups of xpm loader - possibly mem overruns fixed.
fix pixel buffer overflow on loading corrupt/malformed/bad xpm's
and the xpm loader also leaked!
add asm/page.h back in. doesn't hurt on x86
same as previous commit.
more commit....
fix unpremul - broken :(
fix eet load of premul images - need to check they are not "bad" :)
dont unpremul
add sparseness checker :)
check sparseness when doing eet loads.
oooh - lots of cache misses due to new load opts - if load opts have "0"
joses fixes for grads with premul
hmm - broke e17's mouse pointer
remove sub-pixel cursor advance for font rendering
asparagus! (finally after all the radical changes of late)
ooh - freeing a smart multiple times - bad bad bad.
slight cleanups
aspara!
yamaguchi-san's dfb patch
asparagus - we have fixed a lot of things this week, so before the weekend of
been doing instrumentation. i'm cuttong down on startup time - a little here
debian packagnig mofo
i was just playing with adding thread supprot to INTERNALS of evas (ie the
remove test progs and data
remove more
fair enough to be more forgiving of malformed utf8 strings - but the problem
asparagus.
evas utf8 patch broke e17's about box. revert
simon's patch is back - fixed. :) still just makes parsing of tuf8 string
check dimensions for safety
fix fade of solid images in xrender engine
doing some cleanups to xr engine
massively improve xrender scaling to be almost perfect. a few nigglies remain
and fix final off-by-a-bit in the xform.
simply as PLACEHOLDER for a glitz engine. i plan on making it almost
and have glitz disabled for now as its pretty much a skeleton engine.
1. evas gets a pipeline with deferred rendering ability (sometimes faster,
nash patch
fix --enable-ptrhead and poly render symbol - oops
oops - disable threads by default. still experimental.
sli is possible- but not optimal.
a few things.
check for set affinity in pthread - seems to bemissing from debian
disable threads for now - seemed ok. this is a weird thread issue - see mail
threads back on - try this.
add pthread libs etc. to evas-config and pkgconfig stuff
remove thread deadlock! werd!
add version to smart classes.
lotys of changes - mayaswell asparagus - and the start of some of my fm
bring gl engine vaguely back up to snuff.
1. add new api calls to 1. set a "native surface" for image objects. this
and actually call the engine calls!
not complete yet - actually i need help with the fragment shader.
fix segv added in
fragment shader finally actually works. it's unclean and fixme's for seeding
and handle updates. next is to fix the row handling.
fix color modulation for gl surfaces. also fix segvs tat can happen. xrender
fix buffer canvas resizes :)
colorspace work mostly. not complete.
well THAT was just silly. fix yuv texture upload bug - row by row get it right.
and use pixelstorei to do the row skips :)
xrender_x11 engine handles yuv colorspace now - software path, but works.
actually dont delay so long in deleting objects. make it happen sooner
1. fixed some inefficient edje loading - moved checks to the cache code where
patches applied as per emails :)
ooh - and remove leak from callbacks not being freed! fixed!
rotation patches for 180 degrees - thankyoooo! :)
oh dear - when i added load opts - i broke speculative caching. fixed!
i've put back timestamp checks - but evas wont stata file more frequently
chasing somethnig really bizarre... but this works right now.
now limit stats per any 1 file to STAT_GAP (2 seconds).
just look at the comments to know what this is about.
fix nul update on swap
hooray! librsvg leak!. if you try and load a file that is NOT an svg with
ok- disable the close - seems librsvg in some versions is so buggy you can't
pager urgent popup patch - good
cedric's patches.
check if someone was silly enough to try clip themselves
fix up #fidef to be consistent
at least use the right fn call names
yuv call removed if yuv support not built
just spotted a bug - i think its efreet related, but this is to make sure it
patches into cvs we go.
add in the work I did for a 16bit engine - for now, a dormant project until i
use alloca - not malloc :)
extensive delete me handling for evas free's and list walks
gustavo's patch on free an empty/unused evas.
1. abort on invalid stringshared del's - bad stuff and should be trapped.
asparagus - pass distcheck.
remove cpuid stuff - not used. causes some build issues
fix custom file info to be less delicate - not perfectly correct but it wont
1. gustavo';s fuzzy rect implementation for updates. fixed so del's work,
sync xcb
splitrectfuzz of 32 seems fine. havent checked other architectures/cpus and
patch seems to give a bit of a speedup at fuzz = 32 overall.
documentation from brett... tiler can compiler again.
add AC_PATH_X in as well - seems to remove the NONE's being seen?
dont overrun mask buffers!
add an engine idle flush
add mroe ifdle flush
remove a flush - should be empty
ASPARAGUS!
add some notes
save a few bytes by packing structures better... simple savings :)
fix from brett
1. fix a lot of things so they pass make distcheck - so many things have
cedric's sdl patch.
cedric's const patches for evas
sort speedup patch from cedric
0
no need for extra nul byte check - while terminated IF its null. so known
i spyed me a bug that captin'! aye! comments in the code about it - hack
adrunko's precise event patch
was screwing around hunting down bug in edje...
wrong gradient call
gradients dont crash now. draw funny tho :-P
actually - yes - in texture drop not del.
grad fixes from jose
move a chunk of stuff to use evas_smart_class_new() - evas_smart_new() will
fix the cache code - it was basically broken and wouldnt free things from
fix cache accounting
\use direct hash - doesnt strdup the string key. also stringshare the img key
and use timestamp (mtime) for cache
and actively remove images when they are KNOWN to have been removed from disk
cast yet pointers. bad!
extra rgb.txt src
png loading changes to fix greyscale loads apparently on 1.2.19 libpng...
futzing around with tilan's mono font issue - some commented otu debugging -
asparagus!
fix infinite loop!
removbe by list item not data item - faster.
working on optimising software-x11 with the one-buffer persistence idea that
formatting
add a DEL callback - called at the very start of object deleteion.
1. default object size to 0x0
printf--
cedric's sdl patch.
4mb imge cache by default - stupid to have it unlimited.
actually - 0 by default.
resolved bug from leonardo.cunha at openbossa.org
fix unknown buffer content chekc problems
oops - wrong order.
As requested -
static func - cedric patch
patches, caching fixes, zone dynamic creation/deletion handling, stuff.
simple x11 linking
missing -lXrender
e:
only 0 on init if image has alpha
hrrm. 0x0 textures - libGL in new nvidia drivers is most unhappy with! fix.
i've added event flags. this allows objects on top to add flags tyo events
and... set event flags to 0 damnit!
and have the event surfivie between object calls - that was an oops!
probably long overdue - if stringshare add/del get a null string - safely
asparagus! also pass distcheck and have common autofoo init that is
rbdpngn (28):
Generic runtime cpu feature testing. This is not used much yet, and shouldn't
Removed unused variable. 1 line indent fix.
Only perform runtime checks if the cpu features are enabled.
Use the new runtime cpu detection functions to determine the correct drawing
It seems libtool on certain platforms (or possibly certain versions) no longer
Don't access static variables from inline functions.
Altivec code for YUV conversion to ARGB. This option cannot be enabled yet, I
Enable the use of altivec functions. A bit of a dirty hack to detect the
Don't test for altivec if it has been disabled.
Those changes should not have gone to cvs yet.
Enable most features by default, and disable them if needed. CPU specific
Hopefully fix build for solaris.
Ok, that didn't work out well, we'll need to figure out something else for
Fix for a build issue some people have encountered with jpeglib.h not
Renamed a variable to avoid a collision with an imlib2 global.
An instance of ft_lib was missed.
glibtoolize
Handle the case of querying the format when cursor position equals the length
When there is formatting but no text, retrieve the height of the font.
Don't re-declare this variable in the loop.
Fix GL engine for big endian systems.
Fix compile error with disabled cairo engine.
Reduce unnecessary clip recalcs.
Defer marking dirty clippees when clip recalc occurs to avoid recursion.
Create a shared style type enum and padding mapping function.
Avoid allocations on "".
Use a djb2 hash for a better distribution.
Remove warnings by updating to the correct type.
rephorm (9):
put this in HEAD too
fix evas_event_feed_* functions
gotcha rasties.
don't use __blend_tmp before initializing it...
set the name to NULL on _del instead of _free
i assume this was meant as an optimization. however, it wasn't correct for some cases (e.g. rgb = 187, 169, 123)
add some documentation
Patch for line geometry from Cedric BAIL
revert broken fix for polygon geometry
sebastid (55):
Fix format data parsing.
Simplify
Ignore
I like is_set better than is_set_get. It's shorter and looks better and
Disable cairo engine. Doesn't build with latest cairo.
Don't have the CAIRO_CFLAGS as a general include.
Cleanup some headers.
Fix deletion loop in evas_free()
sssh
Silence
Silence
Update functions to match ecore_hash.
Cairo backend builds again. But is dead slow and isn't complete.
Make sure we don't get complaints about cairo engine.
If the version from config has something after x.y.z, drop it.
indent.
Use AC_PATH_GENERIC
Make headers selfcontained.
freetype 9.5.X should be 2.1.7, which is required.
These headers aren't needed here.
Not usable without Evas.h, user should include it first anyway.
Move freetype include along with the others, and use freetype define.
Use right HEAD
Move macros to separate header file.
Remove warning
Add missing header
Use image api right.
Simpler check for 0 stringlength.
Clean warnings.
Don't return value in void func.
Unused variable.
Clean warnings.
Remove strdup
Remove double free
Remove EAPI warnings.
Fix compiler warnings.
Remove the const. It generates a _lot_ of warnings.
Don't remove const. Other code should be fixed instead.
Cast the objects to object list.
Formatting.
Move comment
Fix compiler warnings.
Add const.
\r\n -> \n
Fix formatting
Can someone who knows this code fix this?
space
Add autofoo stuff for svg
Remove unused variables.
Use standard list
Exported module functions must be EAPI
Functions used by modules must be EAPI
Fix typo.
Fix more typo.
Remove unused variables.
shadoi (8):
Rename package to match raster's changes
Add "Provides: libevas" to loosen dependencies.
Disable directfb for now, sid doesn't have a recent enough version.
Standardizing on -bin pacakges for libs.
Include all the bin files
Buh-bye
Include the right file in EXTRA_DIST.
* Run autogen.sh when building CVS snapshots, but not for dist tarballs
shamgar (2):
Cleanup some spacing issues, no extra blank lines in --help this way.
Oops, almost forgot to do this one too.
shorne (7):
Refuse zero length fonts.
Evas on-demand font loading
Make sure to finish initializing sourced fonts
Revert to old fontconfig searching
Add font charmap caching
Add API for getting the available fonts
Fix problem with fonts that cant load completely
sleuth (1):
Minor evas fix for char position usage.
stffrdhrn (5):
FIXME: what do you do when fi is NULL?
Match up function names so gl_xll can work
Add fontconfig to evas for finding fonts
Dont crash on pcf fonts
Evas Fontconfig changes
swielinga (8):
Debian copyright file fix in HEAD replacing Author(s) by Author or Authors.
Big overhaul of a lot of the debian packages
Updated most of the debian packaging. Everything I have missed is out of date,
Added check so that -mcpu=pentiumpro isn't added to the CFLAGS by debian/rules
Fixed a couple of debs in e17/libs, added and updated a couple of .cvsignores.
I Guess this isn't needed (anymore?)
Bump version number of the evas debian package
Updated evas debian packaging.
technikolor (1):
Two minor fixes:
term (9):
Adding yet another spec file.
typo
I was wondering why it worked anyway...
more typos
Whee let's just make every library over! And from cvs! And make it work!
typo (identified by srl).
More cleanups by srl <srl at penguinpowered.com>.
ehco -> echo
One line include addition suggested by Kirby Kuehl <kkuehl at cisco.com>.
thalma (1):
Added checks and fixed built order
tillsan (1):
follow DFB API changes, so the soddy test app can be canceled again
tilman (33):
cleaned up the transformation matrix code. use the proper x macro to convert a double into a matrix component instead of hardcoding the bitshift. added init_transform() which will initialize a matrix to the identity transform.
um, raster, you cannot chain == like that. gcc even told you ;)
don't set the picture transform to the identity - that's a no-op
missed another case of identiy-transform setting
use the proper macros for the picture filter names
avoid rounding errors when setting up the transformation matrix
use double precision for the greater good of humanity
move the filter lookup to its own function - should make the code more readable and makes it easier to play with the filter settings
xrender's filters also kick in if we don't specify a transformation matrix (that's _not_ a bug), so only specify a filter if we're scaling the image
ported the transformation matrix fixes from the xrender_x11 engine to the xrender_xcb engine
ported filter fixes from the xrender_x11 engine to the xrender_xcb engine
don't forward noop move/resize calls to smart objects anymore. code wrapped in ifdefs so you can easily restore the old behaviour if needed.
fixed evas_common_hinting_available so it works properly with both freetype 2.1 and 2.2
don't rely on TT_CONFIG_OPTION_BYTECODE_INTERPRETER in freetype 2.1, because distros suck so much. oh well.
move fontconfig libs to requirements
don't list eet_libs separately in evas.pc -- it's in in requirements already
evas now uses pkg-config to detect EET. i also reworked the autoconf code for that, IMO it now is MUCH cleaner and easier to read than before. if this causes any trouble, please contact me rather than putting weird code in again. thanks :)
use pkg-config to locate EDB, too.
require eet only once
evas-config needs to list EDB_LIBS, too
moved stuff to Libs.private
shorten the freetype check a bit
ugh, removed pkg.m4
tweaked fontconfig check
beautification
added per-engine pc files
oops, forgot them backslashes.
a bunch of constness fixes
missing commas in important places ftl
rework how the html entity escape arrays are stored. this saves a few hundred bytes on the data section of the library, kills off ~300 locations, and makes the so 77K smaller.
Replace another deprecated macro.
When we're drawing text objects, don't apply the horizontal bearing ourselves.
export evas_common_convert_yuv_420p_601_rgba
titan (1):
Remove the comma, as it isn't consistent with the rest of the output.
tsauerbeck (31):
free the freetype handle after we're done with it.
take care of mouse wheel callbacks in _clear() and _cleanup()
fixed evas smart object refcounting
initialize that memory block before reading from it
added evas_common_image_init()/shutdown()
updated
fixed some warnings
added valgrind suppression file for Evas
fixed compilation error when sse was disabled and mmx was enabled
Initialize image data to zero if we're running under Valgrind's control. This means the suppression file is no longer needed.
fixed dist stuff
consistent init/shutdown system
don't try to get the previous character to position 0
remove unneeded include
mingw portability
evas_list_nth_* improvements, step 1
evas_list_nth_* improvements, step 2
got rid of data subdirectory in pkgdatadir. fixed a bunch of warnings
make evas_list_sort more bullet-proof
updated debian files because of recent directory shuffle
fixed a segfault in evas_object_top_get(). by cleaning up the code :)
use valgrind's user requests to tell it that our image data should be treated as being readable
valgrind support requires valgrind 2.4.0 or greater, since 2.2's pkgconfig file seems to be borked
fix printf call
code cleanup
ummm, how did that happen? fix evas_common_convert_rgba_to_24bpp_rgb_888
oops, use the correct type
more warning fixes
some warning fixes
free image cache on shutdown
properly sht down evas etc, so evas_software_x11_test can be used to check for leaks. fixes need to be applied to the other engine tests as well
urandom (1):
s/LIBCAIRO_SVG/LIBSVG_CAIRO/
vapier (16):
allow for setting of env var to prevent running ./configure like the older autogen scripts allowed
remove AC_CANONICAL_TARGET and use $host* variables instead of $target* variables since thats how it works
fix underquoting
add error checking to all autogen scripts
fix whitespacing in a few configure options and remove [] around tests
fix whitespacing
cleanup altivec checks to be more portable
include altivec.h if we have it
touchup help output
re-enable new directfb code since 0.9.2[34] has been released
cut old code
touchup --help output
fix double output for xpm
remove duplicate cairo-svg check
fix up some orphaned context handling in previous commit
we dont actually use any PAGE defines, so dont include asm/page.h
xcomputerman (7):
Bye-bye acconfig.h
- Don't check for C++ compiler if Qtopia isn't requested
This too
Fix-moc-compiler-meta-dependency-distdir-kung-foo
- Fix __am_fastdepCXX bug once and for all. If you don't have a C++ compiler you're just gonna have to edit configure.in by yerself. :)
That shouldn't have gone
Cure apparent symptoms of rasteritis. (compilation errors).
xenith (1):
If cvsignores were beers, I'd be dead by now.
xstasi (3):
changed maintainer, removed changelog autogeneration, fixed build-deps problems which caused errors on building under testing/unstable, and removed CPU optimizations to make it build on non-x86 platforms
Recreated debian/, based on cdbs
redundant line debian/changelog in maintainercleanfiles caused packages not to build
zuluone (1):
corrected the docs slightly - no code changed
-----------------------------------------------------------------------
--
Enlightenment DR17 advanced canvas library
More information about the Pkg-e-commits
mailing list