they form more of a logical group, so nothing lost here, just now its
ALL inside conf_display (like conf_applications actually).
ths does NOT mean we merge every category entirely.
for example (this is kind of a plan):
in input i'd merge key bindings, mouse bindings AND i'd bring over
acpi bindings. edge bindings i'd keep alone for now. interaction and
mouse settings i'd merge.
in windows i'd merge everything except window list and window
remembers. window list i'd merge into the winlist module itself as its
the configuration FOR that module (and then config for it i'd move to
its own config file). window rememebrs i'd keep on its own because its
a complex thing that might want to be totally hidden or re-vamped on
its own.
in menus i'd merge client list menu over to the merged "windows"
module (change its name to Window List Menu too).
in language i'd merge both language and input method setting. both
are related to dealing with multiple languages (input and display).
in look i'd leave wallpaper2, and merge wallpaper, theme, colors, fonts,
startup, icon theme, transitions and scaling. i'd merge merge mouse
cursor look over to the mouse settings + interaction module up in
input (but keep it in the look category). borders i'd merge over to
the "big windows merged module" but keep it in the look category.
in advanced i'd merge performance and engine. leave the rest.
in settings i'd leave it as-is.
in extensions i'd move shelves over to the screen category, but keep
it as a module of its own. pager i'd move to the screen category.
leave mixer and connman where they are. everything i'd keep here
- but i'd be tempted to say all the evry modules should be merged into
a single everything modules. they can keep their entries though.
gadgets i'd move over to the screen category
in files i'd merge file icons and file manager modules. keep 2 conf
entries tho (ie conf_mime joins fileman module). yes - i know e has
file selectors and they use the mime conf too, but to most people they
will just accept the file selector as-is and if they want to configure
icons per file tyope.. well.. load fileman module (can turn off
desktop icons if u want).
-------
why do this? fewer modules to load for e. as such e does spend a
fair bit of its startup time thrashing the disk around loading tonnes
of miniature modules. merging them means less thrashing. there is an
argument to be made that these should even become external processes,
but then we'd need to allow them to remote configure e and thats a
complex beastie in and of itself. we could also load and unload some
modules on the fly. this requires extra features in e17's module
setup, but can be done. worry about this for e18/19 etc. for e17 just
reduce the module count to a saner number (outside of the conf modules
which were the worst here, everything and illume are the next worst.
as above - evry could merge i think. illume vs illume2 cant merge, but
i'd consider merging the toggle modules, blutetooth, indicator and home
modules and then the keyboard and softkey modules (as they occupy the same
screen space basically). so that'd take it to illume, illume2,
illume-home, illume-key
how's that for a plan? who wants to help. this is easy stuff really.
just re-shuffling files and makefile.am content and some module
desktop.in files, and inserting some hooks. in module main setup funcs
and.. fixing e config profiles to not load the removed mods.
SVN revision: 58282
its still a hack. real solution is making e totoally composited and
then the state machine can easily handle it all nigh flawlessly.
SVN revision: 58221
with its own objects directly and expose comp canvas to gadgets/fm and
so on just like container canvas does now. a lot of e's code can be
simplified if we went totally compositing... worry about that for e18,
but for now put in hooks).
SVN revision: 58112
- set focus at the end of deskflip animation
- delay show after first move to not draw at final position first
- move window further out to decrease the chance of having comp
get border_hide event before final position..
SVN revision: 58092
line 2901:
(bd->changes.pos && !bd->changes.size) || (!bd->changes.pos && bd->changes.size)
cant be the case as before (!bd->changes.pos) && (!bd->changes.size) was checked.
line 2935ff: new_client is always false after eval
bd->changes.size/pos is always false after eval
SVN revision: 58089
Hannes Janetzek
Massimo Maiurana
Boris Faure
Carsten Haitzler
Daniel Kolesa
Christopher Michael
Miculcy Brian
Lucas De Marchi
Rui Seabra
Mike Blumenkrantz