| Summary: | [EBUILD] update: app-cdr/sync2cd-1.1.ebuild | ||
|---|---|---|---|
| Product: | Gentoo Linux | Reporter: | Remy Blank <remy> |
| Component: | New packages | Assignee: | App-Backup Team <app-backup> |
| Status: | RESOLVED FIXED | ||
| Severity: | enhancement | CC: | j.habenicht, python |
| Priority: | Normal | Keywords: | EBUILD |
| Version: | unspecified | ||
| Hardware: | All | ||
| OS: | All | ||
| URL: | http://www.calins.ch/software/sync2cd.html | ||
| Whiteboard: | |||
| Package list: | Runtime testing required: | --- | |
| Attachments: |
sync2cd-0.9.ebuild
sync2cd-1.1.ebuild |
||
|
Description
Remy Blank
2005-05-16 15:53:09 UTC
Created attachment 59064 [details]
sync2cd-0.9.ebuild
To be put into app-cdr/sync2cd.
looks good Is it normal that the ebuild has to call python_mod_compile in pkg_postinst to compile the python modules? Wasn't this automatic before? I see at least one other ebuild (gnome-python) calls python_mod_optimize, and python_mod_cleanup in pkg_postrm. Should the latter also be added? Remy, You're right - python_mod_compile in src_install is a bit odd. I'm not sure if anything has changed as I don't maintain many (any?) python modules. I'd look for more examples of dist eclass usage. When I get around to this one I'll consult some of the more python savy devs. Looking at distutils_src_install(), it seems that starting with python-2.3, modules are not compiled through distutils anymore, but later in distutils_pkg_postinst(), where python_mod_optimize() is called with every module in PYTHON_MODNAME. This will compile the modules, but only if they are packages, i.e. subdirectories of /usr/lib/pythonX.Y/site-packages. This ebuild, however, installs just one file at this location. The comment at the beginning of python_mod_compile() seems to indicate that this is the function to call for the job. That's why I added the call in pkg_postinst(). Would it make sense to change distutils_pkg_postinst() to call python_mod_compile() if a module in PYTHON_MODNAME is a file instead of a directory? I could write a patch for that, if it does. python people - any comment on comment #5 Daniel, I noticed you added an ebuild for sync2cd-0.9 to portage, pretty much identical to the old 0.8 ebuild. Did you get any answers to the questions above? I'm always interested in improving my ebuild writing skills ;-) sorry Remy - I'm not that familiar with the internals of python building. I was hoping on of the python herd members could answer your question. No problem. But would it be possible to add the dependency on virtual/eject, as
it is used during restore, and to add the src_test() function so that the unit
tests are run? This would allow catching platform-specific bugs.
RDEPEND="virtual/eject"
src_test() {
cd ${S}/tests
./runTests.py || die "Unit tests failed"
}
Version 1.0 is out. Created attachment 84149 [details]
sync2cd-1.1.ebuild
Ebuild for new version.
I added running of the test suite during the test phase.
Comment on attachment 84149 [details]
sync2cd-1.1.ebuild
Version 1.1 is out. Massively decreased memory usage for large archives.
*** Bug 133105 has been marked as a duplicate of this bug. *** Fixed In CVS |
