[with patch, positive review] parallel testing: sage -tp foo/bar/file.py should assume 1 thread

[mabshoff]

Oops:

./sage -tp -long devel/sage/sage/plot/plot.py
Global iterations: 1
File iterations: 1
TeX files: 0
Usage: sage -t <files or directories>.
For more information, type 'sage -help'.

[gfurnish]

Missing number of threads parameter, invalid.

[gfurnish]

This should be the responsiblity of sage-sage

[jhpalmieri]

Here's a patch. If the first argument after "tp" can't be converted to an integer, this sets it to 1 and assumes that the first argument is part of the list of files. The patch also expands on the error messages, changes the usage warning from "Usage: sage -t" to "Usage: sage -tp", gives a pointer to "sage -advanced" for more help, and adds a corresponding string to the output of "sage -advanced".

[ddrake]

With this patch, when I do something like "sage -tp fjfjfjfjfj", it doesn't tell me that I gave it a bad file:

drake@klee:/var/tmp/sage-4.1.2.alpha2$ ./sage -tp fjfjfj 
Global iterations: 1
File iterations: 1
Using cached timings to run longest doctests first.
Doctesting 0 files 
 
----------------------------------------------------------------------
All tests passed!
Timings have been updated.
Total time for all tests: 0.1 seconds

Also, when it makes an assumption about the number of threads, I'd like it to print that it's using 1 thread. (Mostly for my own reassurance.)

[jhpalmieri]

With this patch, when I do something like "sage -tp fjfjfjfjfj", it doesn't tell me that I gave it a bad file:

I feel as though this belongs in another ticket. For what it's worth, previous authors of the file seem to have made this decision intentionally:

               else:
                    continue # prefer silence to: raise TypeError, "Unknown File %s" % F

Also, when it makes an assumption about the number of threads, I'd like it to print that it's using 1 thread. (Mostly for my own reassurance.)

That's easy enough to change; see the new patch. This version also sets numthreads to be no more than the number of files.

[ddrake]

Replying to jhpalmieri:

With this patch, when I do something like "sage -tp fjfjfjfjfj", it doesn't tell me that I gave it a bad file:

I feel as though this belongs in another ticket. For what it's worth, previous authors of the file seem to have made this decision intentionally:

               else:
                    continue # prefer silence to: raise TypeError, "Unknown File %s" % F

Okay, that does look like another ticket.

Also, when it makes an assumption about the number of threads, I'd like it to print that it's using 1 thread. (Mostly for my own reassurance.)

That's easy enough to change; see the new patch. This version also sets numthreads to be no more than the number of files.

Those look like good changes. Magic 8-ball says...positive review likely.

[ddrake]

Oops, there's a typo: numthreads = min(numthreads, len(files)) should be *after* files is populated; otherwise, a directory with tons of files counts as one file! If you move that bit to line 324 (after the populatefilelist(infiles) stanza), everything works as expected. With that change, I give this patch a positive review. John, after updating your patch, you can change the title to positive review.

Release manager: please check the files COPYING.txt, install, and spkg-install into the sage_scripts hg repo! Right now they're not tracked at all.

[jhpalmieri]

Replying to ddrake:

Oops, there's a typo

Good catch. Fixed in the new patch.

[jhpalmieri]

apply to scripts repository

[mvngu]

Replying to ddrake:

Release manager: please check the files COPYING.txt, install, and spkg-install into the sage_scripts hg repo! Right now they're not tracked at all.

When packaging the Sage source tarball, the files COPYING.txt, install, and spkg-install all end up in the package sage_scripts.x.y.z.spkg. When building from source, those three files end up as

• COPYING.txt becomes SAGE_ROOT/COPYING.txt --- The directory SAGE_ROOT is not tracked:

  [mvngu@sage sage-4.1.2.alpha2]$ hg st
  abort: There is no Mercurial repository here (.hg not found)!
  

• install becomes SAGE_ROOT/spkg/install --- The directory SAGE_ROOT/spkg is not tracked:

  [mvngu@sage spkg]$ pwd
  /scratch/mvngu/release/sage-4.1.2.alpha2/spkg
  [mvngu@sage spkg]$ hg st
  abort: There is no Mercurial repository here (.hg not found)!
  

• spkg-install becomes SAGE_ROOT/local/bin/sage-spkg-install --- The directory SAGE_ROOT/local/bin is indeed tracked.

As it now stand, the changes to spkg-install can be checked in to SAGE_ROOT/local/bin/sage-spkg-install. Changes for the other two files would need to be manually applied to the relevant files.

[ddrake]

Replying to mvngu:

Replying to ddrake:

Release manager: please check the files COPYING.txt, install, and spkg-install into the sage_scripts hg repo! Right now they're not tracked at all.

When packaging the Sage source tarball, the files COPYING.txt, install, and spkg-install all end up in the package sage_scripts.x.y.z.spkg. When building from source, those three files end up as

• COPYING.txt becomes SAGE_ROOT/COPYING.txt --- The directory SAGE_ROOT is not tracked:

[snip]

• install becomes SAGE_ROOT/spkg/install --- The directory SAGE_ROOT/spkg is not tracked:

[snip]

• spkg-install becomes SAGE_ROOT/local/bin/sage-spkg-install --- The directory SAGE_ROOT/local/bin is indeed tracked.

As it now stand, the changes to spkg-install can be checked in to SAGE_ROOT/local/bin/sage-spkg-install. Changes for the other two files would need to be manually applied to the relevant files.

Okay. It does seem weird that basic source files are not version controlled...but whatever. :) For the purposes of this ticket, you just need to merge the patch to sage-ptest.

[mvngu]

Merged in the script repository.

[mvngu]

There is no 4.1.2.alpha3. Sage 4.1.2.alpha3 was William Stein's release for working on making the notebook a standalone package.