On Tue, Mar 09, 2004 at 11:36:20AM -0500, Brian Kelley wrote:
> Jean-Claude Wippler wrote:
>
> >Uh, oh. Dict is used as sequence. Key order changes.
This may be the case, but I can't see where this is a Python-specific
bug then; once a hashed view is established, Mk4py doesn't treat it
any different from a regular view.
> Crap. this needs to be fixed ASAP. I'll download the code now and make
> a patch unless you want to contact Nicholas. I haven't run into this
> before because I tend to just use sequences for appending.
I've adapted the test framework to handle hashed views and sequences
vs. dictionaries (it previously used dictionaries for everything, but
didn't test hashed views), and stuck your test in there. I've traced
it through the Mk4py layer and can't see anything different happening.
If the dictionary order was actually affecting anything, then the
wrong elements would be inserted in the columns; yet this doesn't
happen.
The current state of my work is:
<http://web.sabi.net/temp/metakit-2.4.9.3+njr.tar.gz>
'python setup.py test' will run all the tests, or 'python setup.py
test --test-options=-v --single-test=test_hash' will run just the
hashed view test.
To run the test script under Python 2.3, you need to apply the
test.regrtest patch here:
<https://sourceforge.net/tracker/download.php?group_id=5470&atid=305470&file_id=61798&aid=808210>
OT: While my patch still hasn't been applied, almost 6 months later, I
do notice the underlying issue of absolute vs. relative imports is
being addressed:
<http://www.python.org/peps/pep-0328.html>
--
=Nicholas Riley <[EMAIL PROTECTED]> | <http://www.uiuc.edu/ph/www/njriley>
_____________________________________________
Metakit mailing list - [EMAIL PROTECTED]
http://www.equi4.com/mailman/listinfo/metakit
