Commits

Show all
Author Commit Message Labels Comments Date
Brandon Rhodes
After thinking more about the danger of making any assumptions about where site.py lives, I have fallen back to an even earlier idea that I had while rewriting the virtualenv site.py: having it just look through the system site.py for a "site.py" file, like Python itself would. The only thing that I think could foil this code is if "site.py" were inside of a zipfile.
Brandon Rhodes
Carl Meyer pointed out that my idiosyncratic attempt to find "site.py" without building a fragile-looking explicit "lib/pythonX.Y" path runs aground on systems like his that have a zip file first in their default sys.path. So I have returned to doing it the normal way.
Brandon Rhodes
As Carl Meyer points out, I should be more aggressive about making sure that packages in the virtualenv appear in sys.path before packages from the system Python. This patch, rather than leaving paths in the crazy order that addsitedir() leaves them in, rearranges them after the fact into a strict order: PYTHONPATH first; then the virtualenv; then the system packages, in whatever order the system would normally have them.
Brandon Rhodes
Bugfix: my path-rewriting algorithm was tramping on PYTHONPATH explicit paths.
Brandon Rhodes
Ran "rebuild-script.py" so that virtualenv carries its new site.py file.
Brandon Rhodes
Bugfix: the first element in sys.path, of course, is not always the base system "lib/pythonX.Y" directory; sometimes it's the directory of a module being run, or of something added to PYTHONPATH. Therefore, I should do exactly what Ian used to do: assume that "lib/pythonX.Y" is the naming scheme on all Python installs, and build the path to the system "site.py" manually.
Brandon Rhodes
Hide sys.exec_prefix from system "site.py", and disable sys.path reorder for the moment.
Brandon Rhodes
Wrote a first version of "site.py" which seems to accomplish most of its goals. The (as usual, rather sensitive) code in setuptools does not seem to be happy with the arrangements yet, so a virtualenv with this "site.py" does not finish building; but a check-in is needed to provide a jumping-off point in debugging the problem.
Jannis Leidel
Moved support-files in a package to get rid of the package_data hack and reliably distribute them
Branches
trunk
Jannis Leidel
Fix issue 6 - Updated setuptools to 0.6c11
Branches
trunk
Jannis Leidel
refs issue 5: Updates URL to bugtracker in docs.
Branches
trunk
Jannis Leidel
Updated activate.sh file in virtualenv.py
Branches
trunk
Jannis Leidel
Fixed oversight in rebuild-script
Branches
trunk
Jannis Leidel
Added manifest template to make sure the docs are added to the source distribution but not installed
Branches
trunk
Jannis Leidel
Moved support scripts out of the way to prevent them from being installed globally
Branches
trunk
Ian Bicking
Added tag 1.3.4 for changeset 4b660247ce56
Branches
trunk
Ian Bicking
update news for 1.3.4
Tags
1.3.4
Branches
trunk
Ian Bicking
news for the last few commits
Branches
trunk
Jannis Leidel
Fixed creating symlinks when using --python; creates a python > pythonX.Y symlink.
Branches
trunk
Jannis Leidel
Fixed #389550 - Makes --relocatable work better with --python.
Branches
trunk
Jannis Leidel
Added tag 1.2 for changeset ffa882669ca9
Branches
trunk
Jannis Leidel
Added tag 1.3 for changeset 737c08e96cfe
Branches
trunk
Jannis Leidel
Added tag 1.3.1 for changeset d1f908558001
Branches
trunk
Jannis Leidel
Added tag 1.3.2 for changeset 12c4b5ee5a7f
Branches
trunk
Jannis Leidel
Added tag 1.3.3 for changeset df38ea235009
Branches
trunk
Jannis Leidel
Fixed #422539 -- virtualenv deactivates correctly on ksh now
Branches
trunk
Ian Bicking
make the doc generation more resilient
Branches
trunk
Ian Bicking
merge changes
Branches
trunk
Ian Bicking
leave out sphinx _build directory
Branches
trunk
Carl Meyer
force global eggs after local site-packages in sys.path
Branches
trunk
  1. Prev
  2. Next