This issue tracker has been migrated to GitHub, and is currently read-only.
For more information, see the GitHub FAQs in the Python's Developer Guide.

Author jvr
Recipients
Date 2002-12-15.14:28:37
SpamBayes Score
Marked as misclassified
Message-id
In-reply-to
Content
Logged In: YES 
user_id=92689

Uploaded a new version. Many changes, but most are fairly
small. Addressed a fairly long list of comments from Guido.

General:
- lots more comments, especially for functions.
- wrapped long lines
- site.py: doesn't check for existence in the file system at
all anymore

zipimport.c:
- fixed a really dumb bug that prevented zipimporter objects
be found in the cache; this improved performance a bit in
the presence of packages. (On my box, zipimports are now
always faster than file system imports, but the difference
is much better for uncompressed zip files.)
- removed zipimporter_new (PyType_GenericNew is all we need)
- replaced all PyMapping_* calls with PyDict_* calls
- reworked module/package finding code: now really doesn't
depend on dir entries to be present in the archive.
Basically went back to how ahlstromjc did it, which makes
much more sense to me now.
- fixed some refcount bugs
- fixed a potential buffer overrun
- reduced the number of fseek() calls somewhat (XXX there's
an unused and unfinished replacement for read_directory() in
zipimport.c that reads the entire directory into memory in
one gulp. This avoids more fseek() calls and avoids many
PyMarshal_Read* calls. Nevertheless I only managed to shave
off about 10% of dir-reading time, which I think makes it
not worth the hassle. If you think this may have more effect
on your platform: please play!)
History
Date User Action Args
2007-08-23 15:18:57adminlinkissue652586 messages
2007-08-23 15:18:57admincreate