Support os.ftruncate on Windows

[zooba]

BPO	23668
Nosy	@vstinner, @tjguk, @zware, @serhiy-storchaka, @zooba
Files	23668_1.diff 23668_2.diff 23668_2.patch 23668_2.patch: Regenerated for Rietveld 23668_3.patch

^{Note: these values reflect the state of the issue at the time it was migrated and might not reflect the current state.}

Show more details

GitHub fields:

assignee = 'https://github.com/zooba'
closed_at = <Date 2015-04-27.05:32:23.111>
created_at = <Date 2015-03-15.01:41:38.618>
labels = ['type-feature', 'OS-windows']
title = 'Support os.ftruncate on Windows'
updated_at = <Date 2015-04-27.05:32:23.110>
user = 'https://github.com/zooba'

bugs.python.org fields:

activity = <Date 2015-04-27.05:32:23.110>
actor = 'steve.dower'
assignee = 'steve.dower'
closed = True
closed_date = <Date 2015-04-27.05:32:23.111>
closer = 'steve.dower'
components = ['Windows']
creation = <Date 2015-03-15.01:41:38.618>
creator = 'steve.dower'
dependencies = []
files = ['38488', '38499', '38500', '38505', '38614']
hgrepos = []
issue_num = 23668
keywords = ['patch']
message_count = 18.0
messages = ['238113', '238135', '238144', '238149', '238152', '238153', '238184', '238185', '238214', '238752', '238841', '238843', '240003', '240007', '240014', '240385', '240537', '240566']
nosy_count = 6.0
nosy_names = ['vstinner', 'tim.golden', 'python-dev', 'zach.ware', 'serhiy.storchaka', 'steve.dower']
pr_nums = []
priority = 'normal'
resolution = 'fixed'
stage = 'resolved'
status = 'closed'
superseder = None
type = 'enhancement'
url = 'https://bugs.python.org/issue23668'
versions = ['Python 3.5']

[zooba]

With _chsize_s (which supports 64-bit sizes, unlike _chsize), this seems fairly trivial to do, but I'll put a patch up for it in case there's something I've missed.

[serhiy-storchaka]

Rietveld doesn't accept patches in git format.

[zooba]

It normally does... I'll regenerate the patch when I get a chance later today.

[zooba]

Regenerated the patch file. Rietveld may not have liked that the parent changeset in the previous patch doesn't exist (I pulled this out of my patch queue).

[zooba]

Hmm... doesn't even know that the issue has been changed. Reuploading with a different extension.

[zooba]

I get a 405 error if I try and upload the patch on http://bugs.python.org/review/23668/

My other patches from last night worked fine, so maybe Rietveld doesn't like this issue for some reason?

[serhiy-storchaka]

At first glance the patch looks good, but I did not test it. Are there tests for truncate() with large files (> 4 GiB)?

ftruncate() is used also im mmap.resize(). Do you want to provide a patch also for mmap?

[vstinner]

I reviewed 23668_2.patch.

[zooba]

It looks like mmap uses pure Win32 APIs for the Windows implementation, and so _chsize_s isn't necessary. There is a comment "todo: need ... a 'chsize' analog" in the file, but I'm not actually sure what that means.

I've made some minor changes from the review, but I had a question about possibly defining HAVE_[F]TRUNCATE in *.c rather than *.h. See the review, and I'll update the patch once we've figured that out.

[zooba]

Updated the patch, since there's been a lot of checkins.

I also removed the pyconfig.h changes and updated the #ifdef in posixmodule.c to enable truncate/ftruncate and define PATH_HAVE_FTRUNCATE.

And I know in the last review I said I'd switch to _Py_wopen(), but if I do that there's no way to avoid passing _O_CREAT, so I opted to stick with _wopen() and pass _O_NOINHERIT instead.

Hopefully Reitveld handles this patch file. I'm sure I'm not doing anything differently from normal...

[vstinner]

Is _chsize_s() available on all Windows versions and all Visual Studio
(msvcrt) versions?
Le 21 mars 2015 04:26, "Steve Dower" <report@bugs.python.org> a écrit :

Steve Dower added the comment:

Updated the patch, since there's been a lot of checkins.

I also removed the pyconfig.h changes and updated the #ifdef in
posixmodule.c to enable truncate/ftruncate and define PATH_HAVE_FTRUNCATE.

And I know in the last review I said I'd switch to _Py_wopen(), but if I
do that there's no way to avoid passing _O_CREAT, so I opted to stick with
_wopen() and pass _O_NOINHERIT instead.

Hopefully Reitveld handles this patch file. I'm sure I'm not doing
anything differently from normal...

----------
Added file: http://bugs.python.org/file38614/23668_3.patch

---

Python tracker <report@bugs.python.org>
<http://bugs.python.org/issue23668\>

---