What's missing for 4.0

[Nicolai Haehnle]

On Wed, Sep 16, 2009 at 4:08 PM, David Nolden <zwabel at googlemail.com> wrote:
> Once again, I think it would be good to re-evaluate what features we need to
> implement before we can release KDevelop 4.0.
[...]
> What's your list?

As a user who has recently contributed one or two patches, my
impression is that the most important thing is to get a stable, i.e.
non-crashing release out. Oh, and fix the website (it's horrible,
seriously).

To go a bit more into detail: I use KDevelop for my Mesa development.
It works pretty well as an advanced code editor with code source code
navigation and auto-completion.

My biggest gripes, roughly in order of seriousness:

1. The debugger is essentially unusable, even for local debugging (and
I really need remote debugging facilities). I realize that that's
being worked on.

2. I would love to have an indentation auto-sensing feature. Different
parts of Mesa unfortunately use different indentation styles, and it
would be awesome if this were detected automatically. Maybe I could
somehow configure katepart to do what I want, but the real solution
would be to have the indentation logic sniff a source file on load,
and just guess the correct indentation setting. It should be possible
to implement this so that it works for 99.99% of source files out
there, which would really make life easier. Anyway, this is not
something that should delay a 4.0 release.

3. I'm not particularly happy with VCS integration. Not that there are
particular bugs, it's just the UI I have discovered so far just seems
wrong for my Git workflow. To be honest, I haven't investigated this a
lot and just sticked to the CLI after a very brief look. This is
definitely not something that should delay a 4.0 release.

4. Build system integration is problematic. I blame this on Mesa's
crazy Makefile system, and so far I've been able to reasonably work
around it using configuration options and so on. Again, this is
definitely not something that should delay a 4.0 release.

5. Some C-style constructs still seem to be incompletely supported in
the parser/duchain. In particular, struct initializers don't work
correctly. That's not exactly critical though, and it's something I
might work on if I find the time.

cu,
Nicolai