[PATCH RFA] process/thread/lwp identifier mega-patch
Andrew Cagney
ac131313@cygnus.com
Fri Nov 24 03:15:00 GMT 2000
Kevin Buettner wrote:
>
> On Nov 21, 5:23am, Andrew Cagney wrote:
>
> > > Comments?
> > >
> > > I need approval on this from Andrew Cagney. It'd be nice though for
> > > the other affected maintainers (nearly everyone) to chime in though,
> > > particularly if you have serious objections...
> >
> > As a concept it is approved and most definitly welcome.
> >
> > We just need to figure out a way of ensuring that the change doesn't
> > de-stablize things too much. One thought is to cut a branch, apply the
> > change there and then release that as a snap. People can then
> > download/test it allowing you to refine the change before committing it
> > to the trunk.
>
> I'm willing to do this, but it will only be of value if people
> actually download/test from this branch. Also, I think it would be
> preferable for the branch to be relatively short-lived. (Say two to
> three weeks?)
Yes, you definitly do not want a branch to hang around like a bad smell
:-) I was thinking of something like a week! Just long enough for
people to test your changes and submit any needed updates.
The reason for suggesing that you use a branch is that it would allow
you to track the needed changes more effectively. If other people make
use of it (eg people testing threads) then it is a bonus.
> > As a really bad tempoary hack, would the macro:
> > #define inferior_pid PIDGET (inferior_ptid)
> > allow you to break the changes down into a few smaller steps?
>
> I don't think this will work since inferior_pid needs to be used as an
> lvalue.
> I'll study my patches though to see if there might be a way to break
> the changes up into smaller steps.
Yes. It may ease the pain.
Andrew
More information about the Gdb-patches
mailing list