lp:ctwm

Branch merges

Related source package recipes

No recipes using this branch. (?)

Create packaging recipe

Related snap packages

Related bugs

Related blueprints

691. By Olaf Seibert on 2021-10-02

Implement the EWMH _NET_CLOSE_WINDOW message.

690. By Matthew Fuller on 2021-07-15

We're already UTF-8'ing elsewhere in the file, so use it in a couple
names. Better than the wrong letter, and WAY better than HTML
entities...

689. By Matthew Fuller on 2021-07-14

Do various non-functional cleanups.

Primarily this clears out all the remaining things clang-analyzer bitches
about (apart from stuff in lex generated code, which ain't my problem).
These basically consist of:

  - Dead store warnings. Some of these are obviously eliminatable;
    others are _correct_, but have instead been silenced because they're
    intentionally dead stores for defensive reasons.

  - Null pointer derefs. Some of these I'm pretty sure are impossible
    due to whole-program invariants, others are probably impossible due
    to how X works.

I'm fairly certain none of them indicate any real problems; if they did,
they'd be manifesting as segfaults, and we'd be hearing all about that.
So this is mostly just about clearing out the list as far as possible to
make it easier to find any meaningful ones that show up in the future.

Also tweak cmake config a bit to get useful compile_commands.json back
for dev use.

688. By Matthew Fuller on 2021-07-14

Go ahead and use a proper list joining technique on cmake versions
that can do it, instead of just always falling back to the stupid hack
that doesn't immediately fail until we can bump the minimum past that.

687. By Matthew Fuller on 2021-07-12

Fixup yacc build after previous rev.

Return to maintaining YFLAGS as a list (which is probably more
sensible anyway), and just stringifying in the stupid available way
for bison. Leave a note about the less stupid way, for use at a time
when we can bump cmake up to only 3 years old.

686. By Matthew Fuller on 2021-07-12

Update minimum cmake version to 3.6.

Our original 2.6 minimum was already a little speculative, as the initial
implementation was done with 2.8. Older versions are rarely tested
anyway, and in the course of this update I came across syntax requiring
3.5 that's been there a while, so versions older than that probably
haven't been tried in years anyway.

3.6 is available in EPEL for CentOS 6, which is out of any hint of
support upstream, and was replaced 7 years ago by CentOS 7 which is
itself already a year out of full updates. This is probably an
excessively conservative lower bound even for us, but it's a step.

See also some of LLVM's discussion about their level bumps last year,
e.g. https://lists.llvm.org/pipermail/llvm-dev/2020-March/140349.html
(currently requiring 3.13.4).

685. By Matthew Fuller on 2021-07-11

Redo figuring feature test macros to be more generic.

Previous code was a hardcoded "Does this look like glibc? Then it
needs this stuff added". However, current Illumos (at least with gcc)
also needs some defs added. So rework it to be more symptomatic; take
the funcs we generally see falling into the cracks, and see which
extra defs may be needed for them.

684. By Matthew Fuller on 2021-07-09

Use a make var for the cmake to run, to make it easier to run it under
a different name.

683. By Michael van Elst <email address hidden> on 2021-06-13

Use SizeFont's height instead of the text's height for figuring the
size of the window for RandomPlacement=off of windows. This seems to
be a slightly less bad guess in practice, ignoring all the other
b0rkedness in how this window is used in its various guises.

682. By Matthew Fuller on 2021-03-10

Escape early from UnlinkWindowFromRing() if the window isn't on the
ring. Bad things happen if you proceed to the assert()'s in that
case...