[R-pkg-devel] CRAN Debian test machine

[@vi@e@gross m@iii@g oii gm@ii@com]

This may be a dumb question but are the tests meant to test actual resource usage beyond some levels OR meant to allow the existing machines to handle the load?

The former case might measure not elapsed time on a busy machine, but the time-shared time that remains about the same even on a busy machine or the number of bytes transferred over a network rather than how long it took on the clock.

Of course, measuring such things can add an extra layer of overhead so we may be stuck with what is being used, and it can be random whether a tests fails if done during some rush hour versus at 3 AM.



-----Original Message-----
From: R-package-devel <r-package-devel-bounces using r-project.org> On Behalf Of Dirk Eddelbuettel
Sent: Saturday, May 16, 2026 3:40 PM
To: Roy Mendelssohn - NOAA Federal <roy.mendelssohn using noaa.gov>
Cc: List r-package-devel <r-package-devel using r-project.org>
Subject: Re: [R-pkg-devel] CRAN Debian test machine


On 16 May 2026 at 12:12, Roy Mendelssohn - NOAA Federal wrote:
| yes exactly - whatever is used by CRAN for the automated pre-test on submission.  Not a complaint.  Just would like not to have to keep resubmitting to CRAN because the Debian pre-tests fail.  Would prefer to catch that before submission.  If not, that is fine,  I will just do resubmissions

I mentioned that CRAN does not provide, say, Docker containers.  That said,
rocker/r-base (Debian) and rocker/r2u (Ubuntu) are close. You could use them.

>From an "ideally someone else should take care of it" angle, your best bet is
probably the Linux builds at r-universe which happen to be on Debian and use
similar sets of packages. But again, the exact parameterisation is not the
one CRAN uses for reasons already mentioned. There are a number of tutorials
about how to get going with r-universe, it is very much worth it.

Completely avoiding rejection is hard. I happened to have had issues for the
last few months where the Vienna (Debian) machine was more stringent on URL
checking in pdfs too, and it took some effort to discover that ... the very
variant of qpdf I am using on Ubuntu 25.10 had a change in its command-line
arguments (since reversed, it seems) that influenced the URL extraction.
Sadly even when we try our hardest given the current setup we still do not
come perfectly close. C'est la vie.

| > On May 16, 2026, at 12:06 PM, John Harrold <john.m.harrold using gmail.com> wrote:
| > I think this is the question:
| > 
| > Is there a test machine that is known to have about the same speed and bandwidth as the Debian test machine.

Even if it existed, how would you account for variable, unobservable, and
presumably significant, load on the CRAN machine?  

Dirk

-- 
Dirk Eddelbuettel | edd using debian.org | http://dirk.eddelbuettel.com

Support my Tour de Shore 2026 ride benefiting Maywood Fine Arts! More info at
https://dirk.eddelbuettel.com/blog/2026/04/03#sponsor_tour_de_shore_2026

______________________________________________
R-package-devel using r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-package-devel