Hi, I was trying to find if it was discussed before, but didn't find anything -- what is your view on spkg dependencies?
It is not that bad so far, but still I need to remember in which order to install all my aditional software, e.g. cmake first, then the fortran package, then my wrappers, it'd be convenient if the user could just do "sage -i qsnake" and it would do the right thing. I also spent couple hours investigating some package managers that handle dependencies and don't require root access. I quite liked: http://0install.net/ There are packages in almost all distributions, but then I tried to install it on our department cluster: http://0install.net/install-source.html and it utterly failed: $ python setup.py install --home ~ --install-data ~/.local running install error: invalid Python installation: unable to open /usr/lib/python2.3/config/Makefile (No such file or directory) Yes, I know there is some old python. So that's a show stopper. Ideally, I would like to bootstrap like Sage, from nothing. E.g. the user would download a small tarball, that would contain basically just a python based package manager + python itself (if it's not installed on the system). And then he could install and remove any packages he wants, they would download & install into ~/.cache or something. So it would work like a source distribution, that runs everywhere. The advantage is that libraries like numpy, lapack, blas, etc. would get installed only once and all the upgrades would just download a small thing. The Sage current approach is to download and compile everything all over again. But I don't want to start anything new, so I am just curious about Sage plans in the future about this. Basically, what I need from Sage is atlas, lapack, numpy, scipy, python and then the possibility to install all my additional packages. Another thing --- I'd like to create some repository with my packages, so that people can just "sage -i" install them, without having to first wget all the spkg and install them manually. So I thought I would get my packages to sage experimental, but is there any procedure for that? I know that all of this is reinventing the wheel and basically doing what linux distributions are doing, but Sage imho is a distribution -- a source distribution that runs everywhere and actually compiles --- well, it's true that each time I tried to compile sage on some cluster (2x so far), it failed :), but I think I am an exception, since I used some older g++, or some other stuff was broken. Ondrej --~--~---------~--~----~------------~-------~--~----~ To post to this group, send email to sage-devel@googlegroups.com To unsubscribe from this group, send email to sage-devel-unsubscr...@googlegroups.com For more options, visit this group at http://groups.google.com/group/sage-devel URLs: http://www.sagemath.org -~----------~----~----~----~------~----~------~--~---
