Neil Conway <neilc@samurai.com> writes:
> (BTW, this *particular* case may be a bad situation to worry about
> phased-obsolescence, but in general I think it's worth paying more
> attention to reducing backward-compatibility headaches for users...)
Indeed, we've been beat about the head and shoulders on that point
often enough. But you can easily bring a development effort to a
standstill by spending all your time on such matters rather than
productive new work. I think the secret is to understand where it's
worth spending effort on backwards-compatibility issues, and where not.
I can't claim to have a real good track record on making the right
choices...
regards, tom lane