On Oct 17, 2020, at 04:56, Ken Cunningham wrote:
>> It's correct that I would rather people didn't blacklist compilers that are
>> capable of building a port. It wastes user and buildbot time and wears out
>> the buildbot SSDs unnecessarily.
> Half (more?) of the tickets on MP are from trying to build port X with
> ancient xcode clang Y.
I don't know that it's that many.
> When those occur, the simple fix of bumping the compiler is generally
> criticized. An often complicated project of rewriting the c++ to make old
> clangs happy then ensues...but I digress. I won't personally try to do those
> kinds of fixes any longer, if bumping the compiler fixes it.
That's fine. If a compiler cannot build a port, and you can fix it either by
changing the code so that it works with that compiler or by blacklisting that
compiler, that's fine. In simple cases I would rather see the code fixed, but
either is fine.
But just to reiterate: I would rather people didn't blacklist compilers that
are capable of building a port. I said this to confirm your previous statement,
"Ryan has made it clear he does not want every build to be forced to a macports
clang compiler". If /usr/bin/clang can build a port, there is no benefit to
anyone in blacklisting it.
