[v3] debug macro unifcation

Benjamin Kosnik benjamin.kosnik@gmail.com
Fri Sep 8 08:13:00 GMT 2006

> Is there a reason why you are applying this during regression only
> fixes?  We have been release branch rules for about a month or longer
> now and libstdc++ still is getting changes which seems a little big for
> a release branch.

There is no release branch. 

pb_ds is new code, and I found errors in
the debug parts of it. I don't want to have a release of new code
that has known API issues, and then have to support it. The patch was
larger than I expected but mostly mechanical, and to extensions which
have historically been allowed more leeway.

The libstdc++ testresults indicate to me that we're in good shape
across all primary hosts. (And have been for months.) I'm not worried
about the branch to gcc-4_2-branch, quality wise.

Part of the library release process is carefully looking at our
supported interfaces and removing bits that shouldn't be in the next
release. I consider these kinds of patches fair game for stage 3.


More information about the Libstdc++ mailing list