On Tue, Aug 3, 2010 at 11:55 AM, Dr. David Kirkby <[email protected]> wrote: > On 08/ 3/10 06:11 PM, William Stein wrote: >> >> Hi, >> >> I would like to suggest an alternative approach to this problem of >> "supported platforms". >> How about if whenever somebody runs >> >> make testlong >> >> at the end an email is sent to [email protected] that contains some >> information: >> >> * description of platform (OS + hardware) >> * list of files where tests failed >> * version of sage being tested >> >> A script will then 100% automatically generate a webpage based on the >> above info that will >> describe exactly which platforms are known to work for any given >> version of Sage. Simple >> as that. If we want the list of "supported platforms" for a *given >> release* to go up, all we have >> to do is get somebody somewhere to type "make testlong". >> >> Obviously, there will be an easy way to disable this. Alternatively, >> instead of "make testlong" we could >> just use a different target, e.g., "make testlongreport".
This is exactly what I was going to suggest. I would do a different command. > I see two issues with this. > > 1) Not everyone has their computers set up so that they can send mail from > the command line. In fact, given the totally obscure nature of sendmail, I > would suggest that the number who do are in a minority. > > Most people will send mail from a client such as Thurderbird. > > 2) It is far from unusual for 'make ptestlong' to generate failures, yet > when those same doc tests are run again they pass. I don't know the reason > for this, but it is a fact. Sometimes it might be resource shortages, but I > don't think that all the problem. I've know it to do this on my Sun Blade > 2000 which only has 2 CPUs, yet 8 GB RAM. 8 GB should be enough to run two > tests. > > One could end up with an unnecessarily high number of failures reported We should filter by hardware and user as well as OS, which would shed light on any bias. Actually, I think this would be good information to have--if a platform is flaky I'd like to know (vs one that works all the time for everyone). It wouldn't be too hard to checksum as well. The way I see it is either (1) someone volunteers to step up and test on all those platforms on every release and resolve any issues that come up or (2) we have a long list of what platforms have been supported in the past, which is a good (though not perfect) indication of what things will be like in the future, and the published information is always up to date and accurate. Of course the release manager should make a good faith effort at (1), but I'm not sure how well that works in practice. Perhaps (1) should be a smaller list, clearly indicating that Sage is expected to work on a lot more. - Robert -- To post to this group, send an email to [email protected] To unsubscribe from this group, send an email to [email protected] For more options, visit this group at http://groups.google.com/group/sage-devel URL: http://www.sagemath.org
