On Thu, Oct 3, 2024 at 11:01 PM Matthias Koeppe <matthiaskoe...@gmail.com> wrote: > > On Thursday, October 3, 2024 at 1:18:44 PM UTC-7 Dima Pasechnik wrote: > > Why is it quicker than running ./sage -t src/sage/lfunctions src/sage/rings/ > (perhaps few other should be added) on a full sagelib install? > > > You are missing the point. > > The main point of testing a modularized distribution is not speed but to > protect against modularization regressions. > Testing the modularized distributions is not the same as passing a subset of > files to the doctester.
I am not missing the point. Of course testing each of the distributions is not the same as passing a subset of files to the doctester. We are taking about a particular example - of developing a piece of number-theoretic functionality, and how using only sagemath-number-theory distribution might benefit the developer. It was suggested that it might be quicker with sagemath-number-theory, as it would only test a part of the sagelib. I pointed out that an option to test a part is already here, for long time. And on the other hand, as at some point one would need to test the whole sagelib, it's more work with the modularised system, not less My point is that the benefits of modularisation are unclear, there are more tests to run, more complexity to understand, for seemingly no gain. > And extra complexity to run wider tests, for sagelib, as it's not > installed by default. > > > Also wrong. There are no changes to what is installed by default. Really? So one cannot even build/install sagemath-number-theory without building/installing the whole sagelib? (once again, we are talking about a hypothetical example of someone installing sagemath-number-theory and developing it) And what's the point of all that "modularisation" then? Looks like just extra baggage to be saddled with. And running 10 or 20 times more tests. > For a wider testing, we should not assume that free and generous > GitHub CI allowance we enjoy now is there to stay. [...] > > Fear about suddenly disappearing resources for our CI is misplaced. Really? Do we have a binding, continuing, contract for free provision of CI services certifying to this? Or perhaps a sizeable amount of MSFT stock to be able to pressure for CI availability? :-) Given that lately I have to prove to GutHub that I am entitled to an academic discount (i.e. a free account with a bit more than the default free account) every year, or even more often (and GitHub keeps sending me bills as if it's not aware of my status), this looks like the days of free GitHub might be numbered. > > Moreover, it has no bearing whatsoever on the PR in question, > https://github.com/sagemath/sage/pull/36380, whose specific scope is to add a > new distribution sagemath-brial and to enlarge the existing distributions > sagemath-objects, sagemath-categories. No, why? This thread is a discussion as to sagemath-categories should proceed as outlined in #36380. By the way, it seems that the only way forward with #36380 is for me or someone else to split the brial part off, as you refuse to do so for an unclear reason. Dima -- You received this message because you are subscribed to the Google Groups "sage-devel" group. To unsubscribe from this group and stop receiving emails from it, send an email to sage-devel+unsubscr...@googlegroups.com. To view this discussion on the web visit https://groups.google.com/d/msgid/sage-devel/CAAWYfq0SOzWpm0Sbhg%2BNAL5e99_m3McLnLvh_221w-uTBvAxWg%40mail.gmail.com.
