classification
Title: Preliminary stderr patch
Type: feature request Stage:
Components: Interpreter Core Versions: Python 3.0
process
Status: closed Resolution: accepted
Dependencies: Superseder:
Assigned To: gvanrossum Nosy List: christian.heimes, georg.brandl, gvanrossum, loewis (4)
Priority: high Keywords:

Created on 2007-10-28 13:51 by christian.heimes, last changed 2007-11-02 21:14 by loewis.

Files
File name Uploaded Description Edit Remove
py3k_preliminary_stderr.patch christian.heimes, 2007-10-28 13:51
py3k_preliminary_stderr2.patch christian.heimes, 2007-10-29 10:52
Messages (12)
msg56880 - (view) Author: Christian Heimes (christian.heimes) Date: 2007-10-28 13:51
Here is another patch related to stdio from me. It creates and sets up a
very dumb and easy stderr writer until the new io infrastructure is
initialized. It makes debugging problems in the init phase much easier.
msg56899 - (view) Author: Christian Heimes (christian.heimes) Date: 2007-10-29 10:52
Improved patch
msg56943 - (view) Author: Guido van Rossum (gvanrossum) Date: 2007-10-30 00:29
Good idea.

Are there any places where PyBytes or PyString are written to stderr? 
Or can you get away with just supporting PyUnicode?

It's "dumb", not "dump".

I wouldn't explicitly remove the old stderr printer -- it will be
deleted automatically by the PyDict_SetItemString() call in
PySys_SetObject().  That way you'll still have it when the
PyFile_FromFd() call fails...

This patch doesn't commute with your 'closefd' patch...
msg56955 - (view) Author: Christian Heimes (christian.heimes) Date: 2007-10-30 05:51
Guido van Rossum wrote:
> Are there any places where PyBytes or PyString are written to stderr? 
> Or can you get away with just supporting PyUnicode?

I could have simple copied the write method from _fileio.c and letting
the argument parser do its job instead of fighting with the API for 20
minutes. :

> It's "dumb", not "dump".

I need an IDE with a context aware spell checker :]

> I wouldn't explicitly remove the old stderr printer -- it will be
> deleted automatically by the PyDict_SetItemString() call in
> PySys_SetObject().  That way you'll still have it when the
> PyFile_FromFd() call fails...

I didn't know that. I'll keep it in place.
msg56974 - (view) Author: Guido van Rossum (gvanrossum) Date: 2007-10-30 18:35
I've checked in the patch you added to issue 1329, with one mod: where
you wrote "int fd : 1;" I removed the ": 1" since a field width of one
bit makes no sense here.
msg56977 - (view) Author: Christian Heimes (christian.heimes) Date: 2007-10-30 18:49
Guido van Rossum wrote:
> I've checked in the patch you added to issue 1329, with one mod: where
> you wrote "int fd : 1;" I removed the ": 1" since a field width of one
> bit makes no sense here.

Can you close http://bugs.python.org/issue1771260, too?

Christian
msg57019 - (view) Author: Georg Brandl (georg.brandl) Date: 2007-11-01 17:44
You should be able to close it yourself now :)
msg57026 - (view) Author: Guido van Rossum (gvanrossum) Date: 2007-11-01 18:49
Georg, could you explain somewhere in the tracker docs linked from the
tracker side bar what the options are? I looked into giving Christian
more tracker privileges before, but I didn't know what to type in the
Roles box. IOW I don't know what the allowable roles are, what they
mean, or where to find a list that describes them.

On 11/1/07, Georg Brandl <report@bugs.python.org> wrote:
>
> Georg Brandl added the comment:
>
> You should be able to close it yourself now :)
msg57062 - (view) Author: Georg Brandl (georg.brandl) Date: 2007-11-02 19:38
I've submitted a new bug in the meta tracker.
msg57064 - (view) Author: Martin v. Löwis (loewis) Date: 2007-11-02 20:59
I just created

http://wiki.python.org/moin/TrackerAccessControl

which specifies all permissions in the tracker in detail.
msg57065 - (view) Author: Guido van Rossum (gvanrossum) Date: 2007-11-02 21:04
> http://wiki.python.org/moin/TrackerAccessControl
>
> which specifies all permissions in the tracker in detail.

Is it so that Anonymous < User < Developer < Coordinator? If so, the
role field could just be a dropdown instead of a list of roles...
msg57066 - (view) Author: Martin v. Löwis (loewis) Date: 2007-11-02 21:14
> Guido van Rossum added the comment:
> 
>> http://wiki.python.org/moin/TrackerAccessControl
>>
>> which specifies all permissions in the tracker in detail.
> 
> Is it so that Anonymous < User < Developer < Coordinator? 

Not in roundup per se, no. It is so in this installation.

> If so, the role field could just be a 
> dropdown instead of a list of roles...

I believe a lot of these permissions are redundant. People
holding the Developer role typically also hold the User
role. It might be possible to drop the User role from the
developers, and the User and Developer roles from the Coordinators,
without loss of permissions - but there is a chance that doing
so would break things in case some permission was forgotten.

Regards,
Martin
History
Date User Action Args
2007-11-02 21:14:25loewissetmessages: + msg57066
2007-11-02 21:04:35gvanrossumsetmessages: + msg57065
2007-11-02 20:59:47loewissetnosy: + loewis
messages: + msg57064
2007-11-02 19:38:05georg.brandlsetmessages: + msg57062
2007-11-01 18:49:59gvanrossumsetmessages: + msg57026
2007-11-01 17:44:20georg.brandlsetnosy: + georg.brandl
messages: + msg57019
2007-10-30 18:49:46christian.heimessetmessages: + msg56977
2007-10-30 18:35:10gvanrossumsetstatus: open -> closed
resolution: accepted
messages: + msg56974
2007-10-30 05:51:57christian.heimessetmessages: + msg56955
2007-10-30 00:29:36gvanrossumsetmessages: + msg56943
2007-10-29 20:14:33gvanrossumsetpriority: high
assignee: gvanrossum
2007-10-29 10:52:10christian.heimessetfiles: + py3k_preliminary_stderr2.patch
nosy: + gvanrossum
messages: + msg56899
2007-10-28 13:51:14christian.heimescreate