New port: py26-scientific-devel

[danmichaelo+macports@…]

My portfile is based on the portfile for py26-scientific. I modified the build.cmd to use numpy instead of numeric

build.cmd		${python.bin} setup.py --numpy --no-user-cfg

Some ports depends on py26-scientific, for example py26-ipython and py26-MMTK. I do not now how to let these use py26-scientific-devel instead of py26-scientific. I hope some more experienced macports contributors can comment on this.

[danmichaelo+macports@…]

Whoops, a typo slid in. I meant “instead of py26-scientific”, not “instead of py26-ipython”.

[ryandesign (Ryan Carsten Schmidt)]

Why do we need a devel version of this port?

[danmichaelo+macports@…]

Hi. I'm not sure that we do, but I can explain why I added it. The stable version (2.8) is about two years old, while the unstable version (2.9) is only a few months newer. Since the unstable version is so old, it seems unlikely that a new stable version is right around the corner. Yet, 2.9 has introduced some new features that e.g. the package ​nMOLDYN (which is not yet on MacPorts) depends on. So I had to install it myself, and thought I could make a port for it in the process.

Btw., are there any guidelines for what to do with development versions in general?

[adfernandes (Andrew Fernandes)]

Hmm... someone might want to write a note to the ScientificPython people and let them know that their version numbering is causing problems for package maintainers.

As a busy academic, I really understand the time and effort constraints that they likely are working on. However, the "official" and "unofficial" versioning that they are using really messes things up.

Can you verify that moving from ScientificPython 2.8 to 2.9 is forward compatible? If so, I can update the port to the newest version.

[jmroot (Joshua Root)]

r85603