Bruce Momjian <maillist@candle.pha.pa.us> writes:
> Not really a problem. Most patch problems are either my mucking with it
> to merge it with other changes, or platform-specific problems that the
> tester would never have see. For example, Win32 used EINTR without
> including errno.h.
And, by the same token, whoever does commit a patch might fail to notice
a platform-specific problem that didn't happen to show up on his
platform. You can't really expect things to be 100% tested before they
hit the CVS tree (that's why we have a release cycle).
My experience so far is that the Postgres CVS sources are pretty stable;
most of the time they work, which can't be said for some other projects
I'm involved in :-(. Not that we shouldn't strive to do even better,
but an occasional breakage is going to happen.
regards, tom lane
