Re: Movable initializer lists (C++ N4166)

[David Krauss <potswa@gmail.com>]

> On 2015–05–03, at 11:27 PM, Marc Glisse <marc.glisse@inria.fr> wrote:
> 
> On Sun, 3 May 2015, David Krauss wrote:
> 
>> (Yes, I know that a “compile farm” exists. It appears to be obsolete; perhaps someone could vouch for it?)
> 
> What gave you that impression? It doesn't have a lot of variety, but it has perfectly usable x86(_64) systems, and some very impressive POWER ones.

It’s hard to judge by looking at https://gcc.gnu.org/wiki/CompileFarm because newer and older machines are mixed together in the list, and some of the ones which are still powerful enough (gcc75, gcc76) are nevertheless a few years old. Maybe servers circa 2010 are still faster than my newish, 4x2-core laptop at building GCC, I can’t really estimate.

Checking the list now, it’s less confusing than it was when I last looked in February, which could also be characterized as a decrease in variety. However, this raises the question of machines being taken offline.

Besides that, are some machines overloaded? If I need to use POWER, will there be a learning curve or brittleness as on Darwin? To avoid trial and error whilst wading into the process, I’m just asking for some personal confirmation of suitability for my particular needs: same-day turnaround for clean rebuilds + testsuite validation. Too many days of my life have gone into setting up GCC builds (and then setting up again elsewhere when a problem comes up).