Re: Do not report build errors in bugzilla

[Peter Barada <peter at the-baradas dot com>]

>It has been often said, we do not provide support for Joe Sixpack to
>build glibc.  I get tons of these "compiling or testing  fails for me"
>bugs in bugzilla.  I'm closing them right away since if there would be a
>compile problem in a correct environment it would be noticed by those
>compiling the code every day.  So, either the bug is fixed right away
>(check the mailing list archives) or, and this is the most likely case,
>something is wrong in the environment the would-be reporter uses.  We
>are not helping to chase down these kind of bugs.  If you insist on
>compiling glibc yourself, use up your own time, not ours.  There is
>always the very much preferred method of using a distribution's glibc
>binaries.  If this is not an option for somebody for whatever reason you
>and only you have to debug the problem on your machine.  If it actually
>turns out to be a real bug, report the *real* problem in bugzilla.
>Reporting "test xyz failed" is worthless.  So, don't waste your time and
>ours with this.

Is there a documented procedure(to quote "correct environment") that
are the preferred method to configure/build glibc?  I ask since if I
am trying out a new pattern in a backend for GCC for a target that is
supported by glibc, it would *sure* help both of use if I knew before
that I was building it *exactly* the same way as a "distribution's
glibc binaries" so I know that if I submit these changes to GCC it
won't break the corresponding glibc build.

Then if someone reports "test xyz failed" without indicating they
followed the proper procedure, you can quite rightly tell them to go
pound sand.

If such a "blessed procedure" exists, where can I find it,
*especially* one for building a cross environment?  Does one also
exist for a 'Canadian-Cross'?

The FAQ may have this information buried in it, but not in a manner
that "Joe Sixpack" is going to bother to figure out.

-- 
Peter Barada
peter@the-baradas.com