I have no idea why this should be the case. Will try it on my end and see
what is happening!
On 18 October 2016 at 11:17, Andreas Tille <ti...@debian.org> wrote:
> Hi Leo,
> I've received a bug report which claims that ruffus probably consumes a
> lot of memory in its test suite. While I do not really like to cut from
> the testing quality I wonder whether you are aware of this issue and
> could find a way to reduce the memory consumption of the test suite.
> Kind regards
> On Mon, Oct 17, 2016 at 02:23:50PM +0200, Santiago Vila wrote:
> > Package: src:python-ruffus
> > Version: 2.6.3+dfsg-4
> > Severity: wishlist
> > Hello Andreas and debian-med.
> > I noticed that this package requires 18 GB of RAM to build
> > (as measured by Committed_AS in /proc/meminfo).
> > In theory, there is no limit at all regarding how much RAM a package
> > may need to be built (there is absolutely no policy about that).
> > [ This is the reason why I'm reporting this as wishlist ].
> > In practice, packages like this one represent a real challenge.
> > To put things in perspective: I routinely try to build all source
> > packages generating at least one "Arch: all" binary package.
> > I have collected memory usage for 16500 of those, and currently I can
> > build all of them but this one and "axiom" with only 8 GB of RAM.
> > This makes python-ruffus to be a real "outlier" (in statistical terms).
> > Would be possible to change the build system slightly so that this big
> > amount of memory required to build is reduced significantly?
> > (say, reducing it to a half or a quarter).
> > Note: I think it's not the build itself what uses so much memory but
> > the tests that are performed at the end.
> > Thanks.
> > _______________________________________________
> > Debian-med-packaging mailing list
> > debian-med-packag...@lists.alioth.debian.org
> > http://lists.alioth.debian.org/cgi-bin/mailman/listinfo/