Re: GUB failing

[David Kastrup]

Heikki Tauriainen <address@hidden> writes:

> On Sun, 2013-10-20 at 15:32 +0200, David Kastrup wrote:
>> "Phil Holmes" <address@hidden> writes:
>> 
>> > So will we have to wait for Heikki to provide a fix, or are you able
>> > to fix it yourself?
>> 
>> Pushed a fix to staging.
>
> This patch just keeps on giving trouble...  I'm very sorry for breaking
> things this badly for other people, this certainly never was my
> intention.
>
> (I wasn't aware of the existence of "my_round", although this could've
> again be remedied by just studying more of the existing code, and in
> searching the manpages for a suitable standard library function, I made
> the unfortunate additional mistake of not checking whether "lround" had
> been part of any pre-C++11 standard.  Thanks for fixing this on my
> behalf.)
>
>
> A small question out of curiosity (and ignorance, I'm not familiar with
> the nature of the automatic checks that are made on submitted patches):
> why wasn't this problem caught before the issue was allowed to enter the
> "Patch-push" stage?  (According to the issue page
> <http://code.google.com/p/lilypond/issues/detail?id=3581>, several
> versions of the patchset managed to pass make.  Ideally, this step
> should probably have failed to prevent the patch from proceeding any
> further through the stages...)

"GUB" is our cross-compiling environment (for MacOSX, FreeBSD, Windows,
etc etc) which tends to use quite older compilers/libraries than the
usual Ubuntu which Patchy testing runs on.  So it's actually not all
that rare that a problem surfaces only when we are trying to compile a
complete release.

A normal Patchy run takes perhaps 25minutes on the dedicated machines, a
release compilation about 6 hours.  So we don't do the latter on a whim,
really.

-- 
David Kastrup