Op 18-09-15 om 02:15 schreef Tennessee Leeuwenburg:
Regarding 'boil the ocean' -- I think that's only true when this PEP is taken to be trying to "solve all dependency interactions". In fact, it's more about letting authors handle common, known issues on specific platforms. That's my view anyway. I know if I had this for my packages, just having one or two of these declared would handle a lot of problems.
Perhaps a simple test case would be to get certain packages to work/build/test on travis-ci.org.

I've spend quite some hours this week to get one of my own packages to build there and I didn't succeed. Well, let me rephrase that: I managed to get three packages working without any real problems, but one of them defied my efforts.

It is the typical difficult usecase of a package that needs psycopg2, mapnik, matplotlib, PIL(LOW), numpy, scipy and so. I use buildout with the "syseggrecipe" that can re-use globally installed packages. I could install psycopg2 with pip by first de-activating the standard travis-ci virtualenv. With the new sudo-less infrastructure I could install matplotlib and numpy globally with an apt-get call. My memory is fuzzy right now, but I think I also installed two libraries from within buildout after installing the appropriate header files. Only mapnik failed because I couldn't install some header files and there was no wheel and other alternatives failed, too.

In any case: such a mix of wheel/apt-get/compile with several tools should ideally be stripped down to just one or two methods.

Reinout

--
Reinout van Rees                          http://reinout.vanrees.org/
rein...@vanrees.org                   http://www.nelen-schuurmans.nl/
"Learning history by destroying artifacts is a time-honored atrocity"


_______________________________________________
Distutils-SIG maillist  -  Distutils-SIG@python.org
https://mail.python.org/mailman/listinfo/distutils-sig

Reply via email to