py-matplotlib-basemap Portfile updated to basemap version 1.0.7

[pgf]

Please find in attachment the Portfile for py-matplotlib-basemap updated to basemap version 1.0.7. Tested on my system (Mac OS X Mavericks, XCode 6.1, MacPorts 2.3.3), everything seems OK.

[mf2k (Frank Schima)]

Thanks. ​Per the guidelines, please instead attach a unified diff of the Portfile so we can easily see what changes you are proposing.

[pgf]

unified diff added.

[larryv (Lawrence Velázquez)]

Thanks.

[larryv (Lawrence Velázquez)]

r127314

[pgf]

As far as I know basemap is ready for python 3.x . I'd really like to have it in pyhton 3.x, but unfortunately I'm not enough expert in port file development to try it myself. Is anyone interested?

[larryv (Lawrence Velázquez)]

It doesn’t look like it would be too hard to add a Python 3.4 subport, but at least two of its dependencies needs to have it added first. You should open a new ticket requesting Python 3 support.

[pgf]

Replying to larryv@…:

It doesn’t look like it would be too hard to add a Python 3.4 subport, but at least two of its dependencies needs to have it added first. You should open a new ticket requesting Python 3 support.

I digged a little more into this subject and actually it turned out that the dependency on py-dap, which in turn requires py-httplib2, is no more necessary.
According to the Basemap ChangeLog it was removed in version 1.0.2 .
On the other hand some functionalities require extra packages which are not currently listed as dependencies:

• scipy : required by the 'interp' interpolation routine, provided by port py-scipy
• Python Imaging Library : required by 'warpimage', provided by port py-Pillow (or alternatively the older py-pil)
• OWSLib : required by 'wmsimage', actually not a MacPorts package, but I managed to create a port #45691 .

What about adding these as variants?

variant PIL description {Install Python Imaging Library} {
  depends_lib-append port:py${python.version}-Pillow
}
variant scipy description {Install scipy} {
  depends_lib-append port:py${python.version}-scipy
}

[cdeil (Christoph Deil)]

I've filed a new issue to discuss Python 3 support for basemap: #46665

[petrrr]

In future, please avoid discussing new issues, adding patches, etc. to closed tickets. Such information will usually get lost. Either reopen if resolution is not complete or open a new ticket.

Reopening this to address the dependency issues, which was raised.

[petrrr]

I modified the dependency accordingly in r132168.

[petrrr]

Cc Me!

[petrrr]

I am just reconsidering r132168. These are optional dependencies for specific functionality, which most users may not need, so maybe they should be installed by default?

On the other hand, adding the proposed variants here, IMHO is not very helpful. Having the dependencies in the registry just makes the system more rigid, while without you can just get or remove the functionality at any time by installing uninstalling the respective package.

So what you think, should I just remove the dependencies again?

[Schamschula (Marius Schamschula)]

Replying to petr@…:

I am just reconsidering r132168. These are optional dependencies for specific functionality, which most users may not need, so maybe they should be installed by default?

On the other hand, adding the proposed variants here, IMHO is not very helpful. Having the dependencies in the registry just makes the system more rigid, while without you can just get or remove the functionality at any time by installing uninstalling the respective package.

So what you think, should I just remove the dependencies again?

Great. This update conflicts with my install of gpredict, which is built against py27-pil…

[Schamschula (Marius Schamschula)]

Cc Me!

[petrrr]

Replying to mschamschula@…:

Replying to petr@…:

I am just reconsidering r132168. These are optional dependencies for specific functionality, which most users may not need, so maybe they should be installed by default?

On the other hand, adding the proposed variants here, IMHO is not very helpful. Having the dependencies in the registry just makes the system more rigid, while without you can just get or remove the functionality at any time by installing uninstalling the respective package.

So what you think, should I just remove the dependencies again?

Great. This update conflicts with my install of gpredict, which is built against py27-pil…

Right! This is another issue. I'll look at this later ...

However, this could be resolved with a path dependency as most pil/Pillow dependent port now do. And maybe we should replace py-pil by py-Pillow at some point.

[petrrr]

In r133806 I removed the optional dependencies again. This partly reverts r132168.