Webware for Python

Chuck Esterbrook wrote:
> It seems like we had stability with 0.9 beta 2. Now we don't have
> stability. I appreciate the improvements, but given that 0.9 was first
> released in Jul 14 and changes are still being made which destabilize
> the code, I'm having my doubts that this will ever end.

I tried to make sure that there was always a stable version in the trunk 
at the dates when Mark wanted to make the beta 3 cut. It should have 
been taken place on Tuesday, but probably he was too busy this week.

I also did not make major changes or refactoring in the code; just 
clean-up, fixes, doco and installation (yes I know this can sometimes 
backfire as well).

> Is there any reason we shouldn't dub 0.9 beta 2 as 0.9 final? Further
> improvements could go into 0.9.1.

Maybe we should do that. I fixed a lot of things after 0.9b2 which 
justify making a version of its own.

-- Christoph

On Fri, 2005-09-30 at 02:55 +0200, Christoph Zwerschke wrote:
> > Is there any reason we shouldn't dub 0.9 beta 2 as 0.9 final? Further
> > improvements could go into 0.9.1.

> Maybe we should do that. I fixed a lot of things after 0.9b2 which
> justify making a version of its own.

Chuck's suggestion sounds good to me -- there's nothing stopping us from
releasing 0.9.1 a few days or weeks from now.

peace,
Jason

--
Jason D. Hildebrand
T: 204 775 1212
E: jason@...

On Sep 29, 2005, at 2:23 PM, Chuck Esterbrook wrote:

> Is there any reason we shouldn't dub 0.9 beta 2 as 0.9 final? Further
> improvements could go into 0.9.1.

I agree this is a good idea.

As you rightly assumed, deadlines and contracts got the better of me in 
recent weeks. My next opportunity to make run at the release is the 
afternoon of October 6. If someone else wants to jump on it sooner, 
that is OK by me. Until I hear otherwise, I have pencilled in 10/6, 
13:00 to do the 0.9 RC release.

  - Mark

> Particularly, the tests using 
> unittest.defaultTestLoader.loadTestsFromName have problems loading the 
> tests. I haven't looked into it yet, but it seems to be coming from an 
> incompatible behavior of the Pyhton 2.4 unitest module.

I have to comment myself on that. No, it was not loadTestsFromName 
itself. It only looked like that because it reports no errors why the 
loading of a module failed. I improved the AllTests.py to output such 
errors so that it became clear what the real issues under Windows were.

These were easy to fix, so all tests are now working under Windows and 
Python 2.4 as well. Only one test needs further work, because it relies 
on the fcntl library which is not available under Windows. Maybe I will 
try to find a workaround for that problem, too.

-- Christoph

I fixed two bugs which had been around a long time and resulted in 
broken unit tests for UserKit and TaskKit. All tests (AllTests.py) are 
now working for me under Linux and Python 2.3. I think it would be good 
to ship the next release with working unit tests.

However, I just saw some tests do not run under Windows and Python 2.4.

Particularly, the tests using 
unittest.defaultTestLoader.loadTestsFromName have problems loading the 
tests. I haven't looked into it yet, but it seems to be coming from an 
incompatible behavior of the Pyhton 2.4 unitest module. Can anybody confirm?

Concerning "actions", I really would like to drop the "OldStyleActions" 
setting because it is confusing, and instead support old style actions 
(i.e. the mapping through methodNameForAction()) indefinitely. As I 
said, I see no compatibility, performance or security issues in doing 
so. Anybody objects?

-- Christoph

> Only one test needs further work, because it relies on the fcntl library
 > which is not available under Windows.

I have fixed that as well. All tests are now running smoothly under Unix 
and Windows.

-- Christoph