On Sun, Mar 25, 2018 at 6:53 AM, Dr. David Kirkby <drkir...@gmail.com> wrote: > On 25 March 2018 at 10:03, Vincent Delecroix <20100.delecr...@gmail.com> > wrote: >> >> Apparently Volker does not agree with what was a kind of agreement here >> >> https://trac.sagemath.org/ticket/24903#comment:3 >> https://trac.sagemath.org/ticket/23533#comment:13 >> >> It would be nice that we take a concrete decision about how much we >> support optional packages (and write it in the developer manual). So far >> Maarten, Jeroen and I are in favor of as much support with optional as >> standard. >> And Volker seems to be against. It would be nice to have more >> opinions.
My initial intention with ooptional packages was definitely that they do *not* have as much support as standard. The point of optional packages was that they are packages that are of significantly less importance to most users, so don't need to be installed in every copy of Sage. I just searched for a while (e.g., via google [1] and looking sagemath.org) for the list of optional packages, in order to give some examples of how there is clearly no intention of them working on all platforms, but I couldn't even find the optional package list... However, if I remember correctly, I have added long ago some packages that were OS specific as optional packages. Those can't work on all supported platforms. Experimental is just a place to put packages with absolutely no expectation of them working at all. All that said, these days package systems (and the internet) have gotten massively better, and it is more reasonable to deliver a small core to users, and have them install what they need. I don't have the impresison that *Sage's* package system has necessarily got significantly good though, e.g., as compared to npm, pip, anaconda's, debian, etc., which all have binary options. In summary: this is definitely a question that deserves to be revisted. The answer definitely used to be what Volker seems to think, namely "optional packages are less supported". However, how supported something is, is partly just a function of how much time we have to put into supporting things -- if we have the time now to support all the optional packages better, and our packaging is good enough that it actually works, then maybe we can aspire to make optional packages much better! [1] https://www.google.com/search?q=sage+optional+packages&rlz=1CAZZAF_enUS778US778&oq=sage+optional+packages&aqs=chrome..69i57j69i61l2j0j69i64.2232j0j4&sourceid=chrome&ie=UTF-8 >> >> Vincent >> > As someone who did a lot of development on Sage years ago, my understanding > was that optional packages were meant to work. In much the same way as you > buy a car, and can pay extra for optional features, such as leather seats, > heated seats, alloy wheels, metallic paint etc. > > If something does not work reliably on all supported platforms, then it > should be considered 'experimental' in my opinion. So in my opinion at > least, if something does not work properly on all platforms, there are > currently only two sensible options. > > 1) Fix the problem. > > 2) Downgrade the package to experimental. > > Otherwise you have no distinction between optional and experimental. > > Dave > > -- > 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 post to this group, send email to sage-devel@googlegroups.com. > Visit this group at https://groups.google.com/group/sage-devel. > For more options, visit https://groups.google.com/d/optout. -- William (http://wstein.org) -- 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 post to this group, send email to sage-devel@googlegroups.com. Visit this group at https://groups.google.com/group/sage-devel. For more options, visit https://groups.google.com/d/optout.
