[Patches] [ python-Patches-448227 ] execfile(dir) no longer works

[noreply@sourceforge.net]

Patches item #448227, was opened at 2001-08-05 13:05
You can respond by visiting: 
http://sourceforge.net/tracker/?func=detail&atid=305470&aid=448227&group_id=5470

Category: None
Group: None
Status: Open
Resolution: None
Priority: 2
Submitted By: Titus Brown (titus)
Assigned to: Nobody/Anonymous (nobody)
Summary: execfile(dir) no longer works

Initial Comment:
'execfile(".")' succeeds w/o an error in the current
implementation.  I'm not sure why the _parse itself_
succeeds, but because a directory can be opened by
fopen() for reading, the check in execfile itself
succeeds.

I added in a check (via stat()) to see if the filename
given is a directory.  If it is, execfile raises an
IOError (same as for filename not existing).  I'm
perfectly willing to admit that this may not be the
best solution ;), but it's less nonintuitive than the
current behavior.



----------------------------------------------------------------------

>Comment By: Tim Peters (tim_one)
Date: 2001-08-05 16:08

Message:
Logged In: YES 
user_id=31435

In context, it assumes you read the earlier:

"""
2.3.:
Q: What setting should I use?

A: That is, of course, hard to answer in the general case.
I use the following .cvsrc file:

diff -c
update -d

This defaults diff to context diffs (almost a requirement 
as everything else is harder to read) and tells update to 
automatically checkout new subdirectories.
"""

Nevertheless, you managed to find the only place the FAQ 
doesn't punch you in throat about context diffs <wink>, so 
I changed it as you suggested.  Thanks!

----------------------------------------------------------------------

Comment By: Titus Brown (titus)
Date: 2001-08-05 15:24

Message:
Logged In: YES 
user_id=23486

Section 3.1 of the FAQ:

If you are using CVS (anonymous or developer) you can use
CVS to make the patches for you. Just edit your local copy
and enter the following command:

cvs diff | tee ~/name_of_the_patch.diff

So that should be 'cvs diff -c | ...' ;).


----------------------------------------------------------------------

Comment By: Tim Peters (tim_one)
Date: 2001-08-05 15:15

Message:
Logged In: YES 
user_id=31435

The Python @ SF FAQ already says:

"""
We like context diffs. We grudgingly accept unified diffs. 
Straight ("ed-style") diffs are right out! If you don't 
know how to generate context diffs, you're probably not 
qualified to produce high-quality patches anyway <0.5 
wink>. 
"""

----------------------------------------------------------------------

Comment By: Titus Brown (titus)
Date: 2001-08-05 14:54

Message:
Logged In: YES 
user_id=23486

errno is now correctly set to EISDIR.  I've also modified it
a bit more so it doesn't try to fopen it if the stat fails,
since I can think of no situation in which stat fails & an
fopen will succeed.

New context (-c) diff attached.

N.B. that the Python @ Sourceforge FAQ should be modified to
reflect the desire for -c or -u diffs ;).


----------------------------------------------------------------------

Comment By: Martin v. Löwis (loewis)
Date: 2001-08-05 14:15

Message:
Logged In: YES 
user_id=21627

Can you please submit unified (-u) or context (-c) diffs 
in the future? Plain diffs, without context, are hard to 
integrated once the underlying file changes.

As for the IOError, I think you should make sure you get a 
specific errno code, e.g. EISDIR. Would you be willing to 
revise your patch in that direction?


----------------------------------------------------------------------

You can respond by visiting: 
http://sourceforge.net/tracker/?func=detail&atid=305470&aid=448227&group_id=5470