Re: Bug tracking / Release Quality Assurance

[Nathanael Nerode <neroden at doctormoo dot dyndns dot org>]

Mark Mitchell quoth:

>If you are laboring along on a branch, please take the time to work on
>fixing a few bugs.  If we all fix three or four, we'll be done in
>no time.  Your branch isn't going to be released until we are done with
>this release, so your work is going to see the light of day much sooner
>if you pitch in now.

Looking at the high priority bugs, I figure:

26 c++ bugs (including *all* the rejects-legal bugs)
A number of the bugs in other categories are actually C++ specific as
well.  A number of the bugs in other categories have already been fixed. 
A number of the bugs in other categories seem to be obsolete
or unreproducible.

Most of the remaining non-C++ bugs are target-specific (with a number
that are host-specific).  Then there are the open-ended bugs (such as
'undocumented options' and 'make sure all ada patches to branch get into
mainline').

Unfortunately, this looks like the kind of situation where most of the
bugs are actually quite hard to take on without intricate knowledge of
specific files.  More people working on these bugs will probably not get
them fixed any faster, unless they happen to be the people who are expert
in the correct areas (like the C++ front end).

More people working on bug-fixing is wonderful, of course; I'm just
warning that it may have a significant delay before having any positive
effects, due to people investigating areas of code they're not used to.

--Nathanael