ncm@zembu.com (Nathan Myers) writes:
> This is good news!
> Maybe this process can be formalized. That is, each official release
> migh contain a source file with various "modern" constructs which we
> suspect might break old compilers.
I have no objection to this, if the process *is* formalized --- that
is, we explicitly know and agree to probing for certain obsolescent
constructs in each release. The thing that bothered me about this
was that pgindent was pushing the envelope without any explicit
discussion or advance knowledge.
There's plenty of historical cruft in PG that I'd be glad to get rid
of, if we can satisfy ourselves that it's no longer needed for any
platform of interest. It's "stealth" obsolescence checks that bother
me ;-)
> After a major release, any modern construct that caused no trouble in
> the last release is considered OK to use.
Probably need to allow a couple major releases, considering that we
see lots of people migrating from not-the-last release. But that's a
quibble. My point is we need an explicit debate about the risks and
benefits of each change. Finding out two years later that a broken tool
was doing the experiment without our knowledge is not cool.
regards, tom lane