Re: [Gimp-print-devel] running tests with somewhat limited disk space?

[Greg T. <gd...@le...>]

Robert Krawitz <rl...@al...> writes:

> On Thu, 10 Oct 2019 13:25:05 -0400, Greg Troxel wrote:
>> I just tried to 'make check' 5.3.3.   It failed generating ppds because
>> it filled my filesystem.  I'm using a RAM/VM-backed tmpfs, which is 12G
>> in size.   I remember having to up the ramdisk size earlier, but 12G
>> seems like quite a lot, even in 2019.
>>
>> I wonder if there is any reasonable way to prune what is tested, or
>> generate/check with a smaller working size, or something else?   
>>
>> (Yes, I of course have bigger disks, but I also don't want to put this
>> kind of write workload on SSD; that's a lot of why my pkgsrc builds are
>> in a working directory in the tmpfs.)
>
> This is the CUPS PPD files; they take a little over 12GB.  I could
> probably cut it down a little by testing the global ones separately,
> which would mean 10.5 and 2.4GB respectively, but that wouldn't really
> solve the problem.

That would help me, and make 12G work, vs not, if that's a 20 minute
kind of thing.  But I see the point that 10.5ish vs 12.5ish isn't a gain
worth paying for.

> Unfortunately, refactoring this test would be a pain, and longer term
> PPD files are going away.  If you want to refactor it, go ahead, but I
> don't really want to put a lot of effort into this.

Given that, I'll probably just switch my ramdisk to 14G and see if that
works.  I don't want to work on this either, given that it seems hard
and not important long term.

Thanks for explaining.