optionally turn on old, new, or both for testing purposes using #if's at
the top of the source, but only the new version is turned on by default.
The question is, is an average 0.00007 seconds border find time worth
the memory hit versus an average 0.003 seconds? Keeping in mind that if
you have the pager module running, it adds two border finds, and taskbar
adds thirty for some strange reason. As a bonus e_app_exe_find is also
much quicker.
SVN revision: 26327
Still #if'ed out for now. Try it if you want. Note, #if'ing it in will
run it AS WELL as the old code and print the times taken, so that
comparisons can be made. The new code is the first time reported.
SVN revision: 26323
Thanks to metrics for helping to debug this.
*It's not a white space argument between raster and I, it's an arguement
between our respective editors. His is the only editor that can handle
odd number of spaces indentation, mine keeps trying to convert it to 4
or 8 character tabs. Sorry for the noisy white space stuff in this commit.
SVN revision: 26241
2. fixes to patches to re-enable smooth scale again
3. fixes to make txtblock colors work
4. fixes to edje for premul + grad api changes
5. fixes to ecore to work in premul space
6. fixes to e17 for premul space
7. fixes to emotion for premul
8. rebuild of .edj files as premul requires themes to be rebuilt
9. removed unused fils from cvs (eap and old img files)
SVN revision: 26235
original. Made that all places now have the instances in the original.
Some time during today I managed to break the border matching, and
possibly other searching stuff. Too tired to track it down tonight,
I'll fix it first thing in the morning.
SVN revision: 26220
the E_App, update related aE_Apps, and send trigger change callbacks.
*Have all E_Apps stored in a hash of paths for faster lookup.
*Don't read in all the details of every app at startup, but do scan the
all directory so that we at least know the filenames. Read in details
when needed.
*When searching the non path info of apps, read in data as needed.
Most app things should be faster now. Some things may be slower, and
much more testing / tweaking will be done.
SVN revision: 26213
issues. Changing the NO_APP_LIST define to 0 will put things back the
way they where. This reuses a list that was sitting there doing
nothing, and avoids adding things onto it that where already there, thus
slowing down searches.
SVN revision: 25863
search it and timestamp it.
* App dialog now does all its own .order file mangling. This is
temporary until e_app is back up to scratch following the cachectomy.
SVN revision: 25857
traces of it, I'll find them. Everything complies and I've been running
this version.
I still feel there are some bugs out there that where hidden by the
existance of the cache. Time to stomp those before hiding them again
with the new cache code.
SVN revision: 25838
sys actions check error/status etc. logout implemented, and used. need to
have status for reboot/halt and for suspend and hibernate.
SVN revision: 25656
enabled, there is always one reference to each E_App (the one in the
cache), and that hides some sloppiness in the object reference counting.
So that everybody can continue to use E without crashing all the time,
this is turned back on pending a reveiw.
SVN revision: 25510