Update (at long last) on the TODO list.
On Sun, 3 Jun 2001, Brian S. Julin wrote:
> Stuff to do before release-candidate (IMHO):
>
> 1) Modify Changelog of LibGII/LibGGI, check/update other in-tree docs.
Still TD
> 2) Add Grab or Split input function to prevent segfault when closing
> visuals whose inputs have been joined.
GII side committed. GGI changes are a few trivial functions
which I'll commit in a bit when my tree is cleaner. The GII
part really isn't related too much to the above, but could be handy.
> 3) cursor keys in dga seem to be malfunctioning
>
> Actually, display-dga as a whole seems nonfunctional,
Haven't looked at this yet.
> 4) Remove KGIcon to an attic, disable fbdev-mga on MGA1064
Not removed yet. Andy did commit changes to fix fbdev-mga on G400, though.
> 5) svgalib problems:
> target doesn't seem do frames
smoke says this just isn't implemented as far as he can see, and would
be a chore to implement; we should ensure that mode requests with
frames > 1 actually fail, rather than pretending to work.
> hoses console on exit on Christoph's system
No longer does so with a newer svgalib.
> won't vc-switch back to X afterward
> doesn't work with ggiteleserver
Have not figured out why these two happen yet.
> 6) Running -inroot under 8bpp X stops other apps from running
This is my main reason for writing this update as the is the last
really sticky issue; The above problem can be fixed by adding an
XSetWindowColormap to put the value gotten from XDefaultColormap
back into the root window when done, but raises the question -- is
it the right thing to do to entirely change the ID of the root window's
colormap as we now do, or should we instead add our colors to the default
colormap and then delete them back out when we are done? This I
ask considering the effect on other X apps including GGI apps. Opinions?
> 7) On matroxfb, perhaps all fbdev, when vc is switched away/back
> displayed frame is always frame 0.
Fixed/committed for generic fbdev, not verified that everything is
good with fbdev-mga. Note to testers -- most apps do not redraw
when they receive an expose event, like they should, and that is
the app's shortfalling. As a result, the fbcon or other GGI/svgalib/etc
apps will corrupt the data in frame1, perhaps even later frames
depending on what resolutions are set on the other vc's
We can only do what is practical in the linux vc system environment;
backbuffering would be needed if we wanted to restore the image;
and that would be prohibitively expensive.
> Stuff to do after release-candidate before 2.0 final release:
>
> 1) Unify docs, decide doc website build policy
It seems docs are ahead of this schedule -- will the new material
generate all the docs we currently have? If so, then maybe we should
delete the old files now?
Other things done: Christoph cleaned up some header install issues
and compiler warnings. We no longer install unusable internal debug
headers. A bug resulting from the -physz option was cleaned from
fbdev. Andy added G400 support as noted above.
--
Brian
