On Tue, Jul 5, 2016 at 9:18 PM, leif <[email protected]> wrote: > William Stein wrote: >> On Tue, Jul 5, 2016 at 10:37 AM, leif <[email protected]> wrote: >>> Erik Bray wrote: >>>> That could and probably should be fixed though. In this case just >>>> changing decode('ascii') to decode('latin1') will go a long way (will >>>> result in mojibake for some code pages, but at least won't crash). >>>> Better still is to try utf-8 first and fall back on latin-1 if that >>>> fails. >>> >>> We do not even support spaces in the pathname (a common problem for >>> MacOS X and Windoze users), since that already breaks a couple of >>> upstream packages. >> >> We shouldn't actively discourage people from improving the quality of >> Sage. Pointing out that Sage is broken in other ways is not a >> compelling argument against fixing a build system bug. > > I'm absolutely not against fixing that /where we can/, I'm just saying > we can hardly "fix" each and every package Sage includes, and we sadly > have to live with that, at least for a while. (Though as a disciple of > Unix, I think putting spaces into filenames and especially folder names > is a bad idea in the first place.
All that means is that Unix-based tools don't handle spaces very well. There's no reason in principle spaces aren't appropriate :) (And actually most of them handle spaces just fine if you remember to quote your paths.) > We should just give an /appropriate/ > error message, to the right time, i.e. asap. In this case, Sage's > top-level 'configure' should bail out until we support "non-ASCII" > characters in the pathname of a Sage source tree. Having a Sage > installation reside in such a folder is by the way a related, but > different problem, as long as you don't rebuild packages at least.) I don't think so. You put something like that in and it's going to stay there *FOREVER*, I guarantee it. It will prevent finding such bugs in other packages. I'd rather just fix those upstream packages. Send them to me if you can't be bothered. Yes, that's an invitation. > I was also especially referring to "[7-bit?] ASCII characters", where > rather [a-zA-Z0-9_] was meant. (Python btw. also implies restrictions > on module-, package- and hence also filenames AFAIK.) Yup. Always seem to have to remind people that characters whose codes are above 0x7F are *NOT* ASCII :) Best, Erik -- 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 [email protected]. To post to this group, send email to [email protected]. Visit this group at https://groups.google.com/group/sage-devel. For more options, visit https://groups.google.com/d/optout.
