Re: CALL_EXPR representation changes coming; need patch reviewer(s)

[Sandra Loosemore <sandra at codesourcery dot com>]

Mike Stump wrote:
> On Feb 8, 2007, at 3:27 PM, Sandra Loosemore wrote:
> > Once the patch is approved, we will probably need to ask for a 
> > check-in window for the final merge and commit, as described in
> >
> > http://gcc.gnu.org/ml/gcc/2006-09/msg00454.html
>
> Hum, what rule for mainline were you thinking of?  A no branch merging 
> and no changes to any CALL_EXPR code seem reasonable.  I'd hope that 
> you'd not need a complete lock down on any changes rule.  If you're 
> thinking about a general no changes lock down, I'd counter propose that 
> you just, up-port to mainline, get that reviewed and approved, svn up 
> the tree, resolve conflicts, build, svn ci, svn up and then do one last 
> build.  The cost to others is then just about 0, unless there was some 
> lagging integration work caused by work checked in before your second to 
> last build and the last build.

That is about what I was thinking.  The problem is if there are more conflicts 
introduced during the "build" period between "resolve conflicts" and "svn ci" -- 
especially if we add "test" after "build".  :-)  If you're willing to accept 
that there may be a time period when I'm still cleaning up merge-related 
problems in checked-in code, that's fine, we can do without the check-in window. 
   OTOH, it'll certainly help reach convergence faster if other people can hold 
off on checking in CALL_EXPR-related changes while we're trying to get this 
nailed down.

-Sandra