Re: [wmii] wmii-4 is dead?

"Suraj N. Kurapati" <skurapat_AT_ucsc.edu> writes:

> Hello,
> From what I've learned from regulars in IRC, wmii-4 development has
> lost steam. Some say this was bound to happen because garbeam left
> wmii-4 to pursue dwm. That's fine if you like dwm, but those of us
> still dependent on wmii-3.1 are presently screwed because there
> isn't a stable wmii-4 release yet.

> In case wmii-4 is indeed dead, here is how I would salvage it:

> 1. discard the editable tag-bars stuff

I've already produced a patch to do this.

> 2. discard the "keep stacking mode for this column but show more
> than 1 client" stuff

This hasn't been implemented anyway.

> 3. add use of client ID numbers in the 9p file system, to avoid race
> conditions (as previously discussed in the list)

I've already done this. Note however, that the current fs code has a
much worse race condition whereby invalid pointers may be referenced if
a file stays open after the client or view to which it relates exists.
I'm not entirely sure what the best way to fix this would be. The
easiest might be to use reference counts and a "dead" flag on clients
and views, and then just give an IO error if a filesystem operation is
attempted on a file related to a dead client or view. Better ways to
handle this problem don't necessarily fit well with a "filesystem
model".

> 4. restore swapping, because we cannot use wmii-2 style tiling
> layout without it... heck! even dwm uses tiling layout ;-)

Presumably this is very easy to do (several lines, which can be copied
back from wmii-3); I myself don't really see a use case, but it seems
that you have one.

> 5. add ability to create empty views

This is already possible with wmii-4, although an empty can only exist
while it is current.

-- 
Jeremy Maitin-Shepard