sage --clone: give option to not rebuild the docs

[jhpalmieri]

With the attached patch, passing the argument -N to sage -clone says not to rebuild the reference manual. The patch also uses the option during the clone in the sage-combinat script: there are usually so many patches applied in the queue, it would make more sense to build the manual after applying the patches, not before.

Apply trac_13245-clone-nodocbuild.patch and trac_13245-clone-cython.patch .

[jhpalmieri]

scripts repo

[jhpalmieri]

Here is a different patch; this one just turns off docbuilding in sage-clone altogether.

[leif]

Ping.

Not directly related, but meanwhile sage-clone also regenerates all Cython-generated files, because we don't copy .cython_version, which is fixed by:

sage-clone

@@ -53 +53 @@
 
 cpdir(os.path.abspath('sage/sage'), os.path.abspath(branch + '/sage'))
 
+if os.path.isfile('sage/.cython_version'):
+    print "Copying over hidden Cython version file..."
+    os.link('sage/.cython_version', branch+'/.cython_version')
+
 def copy_dtree(src_dir, dest_dir):
     src_root = os.path.abspath(src_dir)
     dest_root = os.path.abspath(dest_dir)

But there are other issues as well (e.g. rebuilding the fast_callable interpreters, which IMHO isn't necessary either).

[leif]

Replying to leif:

But there are other issues as well (e.g. rebuilding the fast_callable interpreters, which IMHO isn't necessary either).

Oh, forgot: Cython-generated header files (*.h) don't contain a special comment on that on the first line (which is probably an upstream Cython bug), so the following currently doesn't work as expected either:

            if ext in ['.h', '.c', '.cpp']:
                if 'Cython' in open(src + '/' + F).readline():
                    os.link(src + '/' + F, dest + '/' +F)
                    os.utime(dest + '/' +F, None)

(For the moment, we could look for ^#ifndef __PYX_HAVE__sage__ in those, probably even without the sage__.)

[jhpalmieri]

Okay, here is a patch to deal with the .h files and .cython_version. Rebuilding the fast_callable interpreters doesn't take too much time, so I'm not dealing with it right now. Leif, would you like to be listed as an author?

[leif]

Replying to jhpalmieri:

Rebuilding the fast_callable interpreters doesn't take too much time, so I'm not dealing with it right now.

Yes, but it also triggers the rebuild of (currently) at least one other Cython module (besides the newly generated five sage.ext.interpreters.wrapper_*.pyx):

...
Building interpreters for fast_callable
Updating Cython code....
Compiling sage/plot/plot3d/parametric_surface.pyx because it depends on ./sage/ext/interpreters/wrapper_rdf.pxd.
...
building 'sage.plot.plot3d.parametric_surface' extension
...

(I think we could just copy over all files in sage/ext/interpreters/, which is in .hgignore.)

Note sure what or whether doc-rebuilding is triggered (upon next sage --docbuild ...) by rebuilding the interpreters though.

Could you explain why we "touch" almost all manually copied (more precisely, linked) files?

Leif, would you like to be listed as an author?

Does the patch introduce new environment variables? :P

[Haven't tried the patches here yet.]

[leif]

Replying to leif:

Could you explain why we "touch" almost all manually copied (more precisely, linked) files?

Ok, seems like Mercurial is to blame. We could use hg clone -U though instead, and create the working copy by ourselves. (Not tested.)

[leif]

Replying to leif:

Replying to leif:

Could you explain why we "touch" almost all manually copied (more precisely, linked) files?

Ok, seems like Mercurial is to blame. We could use hg clone -U though instead, and create the working copy by ourselves. (Not tested.)

Or change back the modification times of the files (of the working copy) Mercurial created... ;-)

OMG.

[leif]

Replying to leif:

Replying to leif:

Ok, seems like Mercurial is to blame. We could use hg clone -U though instead, and create the working copy by ourselves. (Not tested.)

Or change back the modification times of the files (of the working copy) Mercurial created... ;-)

OMG.

 Thank you, nanny!

(How about making this optional?!?)

[leif]

Since apparently hg archive (meanwhile?) doesn't preserve modification times either, we could use hg clone -U and some cp -pPR of hg manifest magic instead (of plain hg clone).

[jhpalmieri]

Given the upcoming transition to git, I don't think it's worth trying to wrestle with Mercurial very much. I don't want to spend much more time on it, anyway.

[leif]

Replying to jhpalmieri:

Given the upcoming transition to git, I don't think it's worth trying to wrestle with Mercurial very much. I don't want to spend much more time on it, anyway.

Yes, hopefully.

Anyway, the title (and parts of the description) still advertise an option to disable docbuilding, but the currently listed patch just disables it unconditionally. Which way do we want to go?

[jhpalmieri]

I think it makes more sense not to build the docs at all. But I don't use sage --clone myself, I only opened this in response to experiences of various sage-combinat users.