[Python] win32all-142 available

In article <3C368646.5000400 at skippinet.com.au>,
mhammond at skippinet.com.au says...
Got it, installed it and it seems to work. Thanks very much for putting
this up again, Mark - I've been battling with trying to build from the
CVS sources, with some success, but even though I enjoy a challenge as
much as the next man, there is a limit to how much interaction with VC++
I can take, being a loyal Delphi developer <g>.

In article <3C368646.5000400 at skippinet.com.au>,
mhammond at skippinet.com.au says...

Hi Mark,

One small issue so far. I am seeing multiple entries in the popup
attribute window after pressing the '.' key. It appears to be related to
levels of inheritence. An example explains it best:

l = list()
l.+----------
class list2(list):
pass
l = list2()
l.+----------
class list3(list2):
pass
l = list3()
l.+----------
The problem also occurs with the new 2.2 properties, and occurs for
classic classes, new object descended classes, and built-in type
descended classes. The problem does not occur with simple attributes.

<a few minutes later>

I've had a look in the source and have come up with a patch that seems
to fix the problem:

D:\Python22\Pythonwin\pywin\scintilla\view.py (line 407)

#------------------------------------------------------
# Fix to remove duplicate entries
#------------------------------------------------------
# OLD CODE
#------------------------------------------------------
# if hasattr(ob, "__class__"):
# items = items + _get_class_attributes(ob.__class__)
#------------------------------------------------------
# NEW CODE BEGIN
#------------------------------------------------------
allitems = []
allitems.extend(_get_class_attributes(ob.__class__))
for item in allitems:
if not item in items:
items = items + item
#------------------------------------------------------
# NEW CODE END
#------------------------------------------------------

HTH.

In article <MPG.16a13250220c63989896cd at newnews.dircon.co.uk>,
carlca at dircon.co.uk says...

Correction:

#------------------------------------------------------
# Fix to remove duplicate entries
#------------------------------------------------------
# OLD CODE
#------------------------------------------------------
# if hasattr(ob, "__class__"):
# items = items + _get_class_attributes(ob.__class__)
#------------------------------------------------------
# NEW CODE BEGIN
#------------------------------------------------------
if hasattr(ob, "__class__"):
allitems = []
allitems.extend(_get_class_attributes(ob.__class__))
for item in allitems:
if not item in items:
items = items + item
#------------------------------------------------------
# NEW CODE END
#------------------------------------------------------

#------------------------------------------------------
I think you probably mean this.

if hasattr(ob, "__class__"):
for item in _get_class_attributes(ob.__class__):
if item not in items:
items.append(item)

## Jason Orendorff http://www.jorendorff.com/

A different patch has been checked in - thanks!

The Pythonwin browser doesn't really work too well either :)

Mark.

In article <mailman.1010272507.6806.python-list at python.org>,
jason at jorendorff.com says...
Your version is much more elegant, but, I'm sure that you would concede,
functionally identical. In my defense, I was so excited at having found
and fixed this (I am a recently converted Python devotee) that I
neglected to put the code through my usual rigorous 3 day "before
hitting the send button" refactoring session.

:^)

I was just worried about

items = items + item

This is an error, if (as it seems) "items" is a list and "item"
is a string.

## Jason Orendorff http://www.jorendorff.com/

In article <mailman.1010292968.32245.python-list at python.org>,
jason at jorendorff.com says...
You're right of course. The confusing thing was that PythonWin behaved
correctly after applying this "fix". After further investigation I have
found that the code is indeed failing at this point, but the exception
is caught by a try except. The implication of this is that this code
does not make any difference which is indeed what I found after
commenting the code out altogether. Interestingly, I have found that if
I comment out the same block of code in view.py in the 2.1 version of
PythonWin, the code completion fails altogether, which, given that
view.py is identical under PythonWin 2.1 and PythonWin 2.2, suggests a
change in behaviour in the underlying Python.

<quick check later using Python console>

Aha! It seems that under Python 2.2, dir(obj), when applied to a class
returns attributes for ancestor classes whereas in 2.1 it doesn't:

Python 2.2 (#28, Dec 21 2001, 12:21:2
Type "help", "copyright", "credits" o
... a = 23
...
... b = 46
...
['__doc__', '__module__', 'a', 'b']
>>>


ActivePython 2.1.1, build 212 (A
Python 2.1.1 (#20, Jul 26 2001,
Type "copyright", "credits" or "
... a = 23
...
... b = 46
...
['__doc__', '__module__', 'b']
>>>

I assume this is by design, although I cannot find any mention of it in
the 2.2 docs. Comments?

This is new since Python 2.2a3. From NEWS.txt:

"The builtin dir() now returns more information, and sometimes much more,
generally naming all attributes of an object, and all attribtes reachable
from the object via its class, and from its class's base classes, an so on
from them too. ..."

G?nter

In article <3c3869e0$0$22450$6e365a64 at newsreader02.highway.telekom.at>,
g.szolderits at gmx.at says...
Cool. Thanks.

Looks good to me. Lack of win32all was the only thing holding me off
from moving to Python 2.2. Thanks for all the work on this, Mark!

One minor point - for Python 2.2, the "correct" place to install
packages is in <Python Dir>\lib\site-packages (see PEP 250).
win32all-142 still installs directly in <Python Dir>. Any chance of
changing this for the next version?

Paul.

I guess so, yes.

Mark.

Mark,
I'm using your new PythonWin version and noticed the following:
- scripts with the extension "*.pyw" are not recognized
- call-outs are not working for keyboards other than US

Is there any change this will be fixed in the future?
Ruud

Mark Hammond <mhammond at skippinet.com.au> wrote in message news:<3C393995.9080804 at skippinet.com.au>...

Is the source available anywhere so I can build a debug version for testing
?

Thanks
Dave Hawkes

"Mark Hammond" <mhammond at skippinet.com.au> wrote in message
news:3C368646.5000400 at skippinet.com.au...

See the win32all pages at http://starship.python.net/crew/mhammond/win32.
It has pointers to getting source from CVS and building from the source.

--
-- David
--
/-----------------------------------------------------------------------\
\ David Bolen \ E-mail: db3l at fitlinxx.com /
/ 860 Canal Street, Stamford, CT 06902 \ Fax: (203) 316-5150 \
\-----------------------------------------------------------------------/