Mailman 3 October 2002 - Python-Dev

tempfile problems
by Thomas Heller Oct. 15, 2002

Oct. 15, 2002

bdist_wininst currently contains this code: from tempfile import NamedTemporaryFile arc = NamedTemporaryFile(".zip") archive_basename = arc.name[:-4] fullname = self.distribution.get_fullname() arcname = self.make_archive(archive_basename, "zip", root_dir=self.bdist_dir) The corresponding checkin message is this: revision 1.35 date: 2002/08/09 16:37:34; author: gvanrossum; state: Exp; lines: +5 -4 Massive changes from SF 589982 (tempfile.… [View More]

2 1

Missing symbol in sre.py (Python 2.2.2b1)
by Eric Johnson Oct. 14, 2002

Oct. 14, 2002

There appears to still be a missing public symbol definition in the Python 2.2.2b1 version of sre.py The documentation implies that finditer should work and indeed it is defined in the module if sys.hexversion > 0x02020000 however it is not listed in the __all__ module variable. This means that if I use import re I am unable to use re.finditer but if I use import sre then I can use sre.finditer. I thought that the recommended approach was to use import re. … [View More]

4 3

RELEASED: Python 2.2.2 (final)
by Guido van Rossum Oct. 14, 2002

Oct. 14, 2002

Ta-daa! The release of Python 2.2.2 is here. This is a fully backwards compatible bugfix release for Python 2.2.1. Thanks to all who submitted last-minute bug reports and fixes -- and thanks even more to those who didn't wait until the last minute! :-) Python 2.2.2 has a large number of bug fixes in all areas of the interpreter and the library; for a full account see the release notes, available via the link below. With special dispensation from the "bugfixes only" rule, a brand new … [View More]

1 0

Can Zope be built on MacOSX?
by Skip Montanaro Oct. 14, 2002

Oct. 14, 2002

I downloaded Zope 2.6.0b1 today from zope.org and tried to build it with Python from CVS, which failed, then Python 2.2.2 (also from CVS, but which should be between 2.2.2b1 and 2.2.2final), which also failed. Upon closer examination, it seems 2.1.x is required for Zope 2.5, though the doc/INSTALL.txt file says nothing about what Zope 2.6 requires. I then pulled the release21-maint branch of CVS and tried to build it on MacOSX. It failed spectacularly trying to compile Python/dynload_next.c … [View More]

3 2

The -U flag
by Martin v. Loewis Oct. 14, 2002

Oct. 14, 2002

Since Python allows Unicode strings in sys.path, we are making progress on getting -U to work. When entering interactive mode, the stumbling block is _idmap = '' for i in range(256): _idmap = _idmap + chr(i) del i Here, _idmap is initialized with a Unicode string, and the chr(i) results are promoted to Unicode, which eventually causes a UnicodeErorr when you get past 127. The work-around would be to write _idmap = str('') for i in range(256): _idmap = _idmap + chr(i) del i With that, we … [View More]

5 8

RE: [Python-Dev] Python-dev Summary for 2002-10-01 through 2002-1 0-13
by Delaney, Timothy Oct. 14, 2002

Oct. 14, 2002

> From: Brett Cannon [mailto:bac@OCF.Berkeley.EDU] > > .. _frameobject.c: > http://cvs.sourceforge.net/cgi-bin/viewcvs.cgi/python/python/d > ist/src/Objects/frameobject.c Might be worthwhile always including a TinyURL as well as the full URL to deal with wrapping issues (I would suggested a Shorter Link, but Alex Martelli showed me a better one). For example, the above URL gets changed to: http://tinyurl.com/1ynh so the resulting entry would look like .. _frameobject.c: … [View More]

3 2

Re: [Python-Dev] Assign to errno allowed?
by Tim Peters Oct. 14, 2002

Oct. 14, 2002

[Brad Clements] > ... > I would like to remind folks that on some platforms, one cannot > just use "errno = 0". On those platforms calling a function is > required to set errno. Except that errno=0 works fine on any platform with a standard- conforming C implementation, and this isn't even a fuzzy POSIX issue -- it's a requirement of the C standard. We can define piles of macros instead, but I expect that, as the years drag on, people will "forget" to use them.

6 10

Python 2.2.2 release tonight - checkin freeze on branch
by Guido van Rossum Oct. 14, 2002

Oct. 14, 2002

I'd like to do the 2.2.2 release tonight, unless Jack has any unresolved problems. (Jack, mail me ASAP if you do!) Please, no more checkins to the release22-maint branch until the release is done (with the exception of the Mac subtree, and Fred & Tim). I still have to update the Misc/NEWS file. --Guido van Rossum (home page: http://www.python.org/~guido/)

2 1

RE: [Python-Dev] Why does the Windows installer use the shortname formof the install directory?
by Moore, Paul Oct. 14, 2002

Oct. 14, 2002

From: Tim Peters [mailto:tim.one@comcast.net] > [Moore, Paul] > > I just installed Python 2.2.2b1 (replacing ActiveState Python) > > and I notice that the installer uses the shortname form of the > > Python install directory. This appears in sys.path, in file > > associations (and hence in sys.argv[0]), and in lots of other > > places in the registry, not all of which I follow the reasons for. > > > > My Python install directory was C:\Applications\… [View More]Python - not the > > default, but also not containing spaces (which I guess could be a > > problem for some scripts). So the short form should not be > > necessary. I've fixed the problem manually by hacking the > > registry, but it's not something I feel comfortable with... > > I must have missed what "the problem" is. What is the problem? Path > components generally have two distinct names on Windows. That's how > the OS works -- we didn't invent this madness <wink>. "problem" is probably a bit strong. It is user-visible in a few places (notably sys.path) and is arguably difficult to read in those places. But it's an issue of aesthetics rather than anything else. OK, so if I have a sense of aesthetics, I probably shouldn't be using Windows :-) Agreed, it's mainly an OS issue. If Windows didn't do the long/short pathname idiocy, and if the practice of using pathnames with spaces had never been promoted, we wouldn't be having this discussion. (Or if Unix people had got into the habit of actually using the freedom thay have to use pathnames containing all sorts of strange characters, we may be even worse off :-)) > > > Would it be possible for the installer to use the long filename > > form > > Yes, but I won't make that change, because we *used* to do it that way and > it confused newbies endlessly, mostly because of embedded spaces. By > default, the installer suggests an 8.3 name so that nobody can get confused. > Pick a non-8.3 name instead, and two distinct names are a fact of Windows > Life you have to live with. OK. If the other way has been shown to cause real problems, then that's a reason not to do that. Having used ActiveState's distribution until now (which uses the long form, and advises you not to use paths with spaces - I've no idea how it copes if you ignore its advice, though...) I didn't realise that you'd been through this. > > (at least in cases where the install path does not contain spaces)? > > I don't know that the WISE scripting language is capable of making this > distinction, and in the absence of an actual problem I'm not going to spend > time digging into it. It's certainly too late to make this kind of change > in the 2.2 line, so if a volunteer appears who cares enough to address it, > it would have to wait for 2.3. That's fair enough. And it's not major enough that I'm pushing for anyone to waste time on it. I just thought I'd mention it in case it was a simple change to a checkbox, or something. I'm always optimistic on a Friday - I save the cynicism for Mondays... :-) Thanks for taking the time to comment. Paul. [View Less]

6 8

PEP 218 update questions
by A.M. Kuchling Oct. 14, 2002

Oct. 14, 2002

I've just updated PEP 218 to match the sets.py module that's in CVS. The original PEP text had a single Set class, and Set instances froze after their __hash__ was called so you could use a set as a dictionary key. sets.py takes a more straightforward approach, with distinct Set and ImmutableSet classes. Two things in the PEP need to be updated to take this change into account. 1) The long-term proposal in PEP 218 lists a single built-in conversion function, set(iterable). This made … [View More]

5 7