Separate the rankwidth library into a standard package

[mjo]

This is a follow-up to,

​https://groups.google.com/forum/#!topic/sage-devel/HF52TdkH0fw

The author of the rankwidth (rw) package was very helpful in getting the library/executable packaged with autotools. The new release is available here:

​http://pholia.tdi.informatik.uni-frankfurt.de/~philipp/software/rw-0.7.tar.gz

The first big change is that it uses autotools, obviously. So the standard ./configure && make should work. But more importantly, we can now compile the shared library without the executable, and without the resulting dependency on igraph. If you use the default ./configure:

$ ./configure
...
configure: error: Package requirements (igraph >= 0.6) were not met:

No package 'igraph' found

But if you pass the --enable-executable=no flag,

$ ./configure --enable-executable=no

it will succeed without igraph and build the library.

[ncohen]

Have you started turning it into a Sage package?

[mjo]

Replying to ncohen:

Have you started turning it into a Sage package?

Nope, no work on that yet.

[ncohen]

Okay, I'll do it tomorrow then.

[ncohen]

New commits:

​9380024

trac #18681: 'rw' as a package

[git]

Branch pushed to git repo; I updated commit sha1. This was a forced push. New commits:

​687888d

trac #18681: 'rw' as a package

[jdemeyer]

The license is GPL version 2 or later.

[jdemeyer]

For upstream contact, I would mention the page ​http://pholia.tdi.informatik.uni-frankfurt.de/~philipp/software/rw.shtml

[jdemeyer]

These lines should be removed from the .pxd file, since they are not in the API of rw:

subset_t *slots

uint_fast8_t num_vertices

Move this inside the cdef extern from "rankwidth_c/rw.h": block (removing the stuff in quotes and uint_fast8):

    ctypedef int uint_fast8_t
    ctypedef int uint_fast32_t
    ctypedef int subset_t

[jdemeyer]

And this is clearly a private function, this shouldn't be in the pxd either:

cdef void set_am(int i, int j, int val)

[git]

Branch pushed to git repo; I updated commit sha1. New commits:

​8002dc4

trac #18681: Reviewer's comments

[ncohen]

I made all modifications except the one about the licence: where did you see that? I said 'GPL2' in SPKG.txt because that's what the COPYING file contains.

Nathann

[jdemeyer]

Replying to ncohen:

I made all modifications except the one about the licence: where did you see that? I said 'GPL2' in SPKG.txt because that's what the COPYING file contains.

Look in the code (rw.c for example).

Note that Sage is released under GPL version 3 (the only license which is compatible with all standard packages), so if it really was GPL version 2 only, that would be a problem.

[jdemeyer]

Do you really need this stuff in quotes:

"uint_fast8_t"

(same for the other two)

[ncohen]

Yo !

Look in the code (rw.c for example).

Well, that's two contradicting sources. Should we ask upstream to use the same in both places?

Note that Sage is released under GPL version 3 (the only license which is compatible with all standard packages), so if it really was GPL version 2 only, that would be a problem.

Sigh... T_T

Nathann

[git]

Branch pushed to git repo; I updated commit sha1. This was a forced push. New commits:

​f055631

trac #18681: Reviewer's comments

[jdemeyer]

Replying to ncohen:

Well, that's two contradicting sources.

It's not contradicting. The code is what matters. Essentially it says "this code is released until GPL v2 or later, see the file COPYING for the GPL v2 text"

[jdemeyer]

rw should be a dependency of the Sage library in build/deps.

[ncohen]

rw should be a dependency of the Sage library in build/deps.

Arg... True :-/

Can you think of a way to make this automatic?

Nathann

[git]

Branch pushed to git repo; I updated commit sha1. New commits:

​3681e06

trac #18681: rw is a dependency of Sage

[ncohen]

Thanks!