[debbugs-tracker] bug#28023: closed (fix make-temp-file race on local host)

[GNU bug Tracking System]

Your message dated Sun, 13 Aug 2017 00:21:03 -0700
with message-id <address@hidden>
and subject line Re: bug#28023: fix make-temp-file race on local host
has caused the debbugs.gnu.org bug report #28023,
regarding fix make-temp-file race on local host
to be marked as done.

(If you believe you have received this mail in error, please contact
address@hidden)


-- 
28023: http://debbugs.gnu.org/cgi/bugreport.cgi?bug=28023
GNU Bug Tracking System
Contact address@hidden with problems

--- Begin Message --- Subject: fix make-temp-file race on local host Date: Tue, 8 Aug 2017 22:38:05 -0700 User-agent: Mozilla/5.0 (X11; Linux x86_64; rv:52.0) Gecko/20100101 Thunderbird/52.2.1

Tags: patch
Severity: important

I plan to install the attached patch in master soon, and am sending it for review to bug-gnu-emacs first in case there's some problem with it on MS-Windows. Although this patch does not fix all instances of the race in Gnu Emacs (notably, Tramp access is still affected) it's a good start.

0001-Fix-make-temp-file-race-on-local-files.patch
Description: Text Data

--- End Message ---

> --- Begin Message ---
>
>
>
> Subject: 
>
> Re: bug#28023: fix make-temp-file race on local host
>
>
>
>
> Date: 
>
> Sun, 13 Aug 2017 00:21:03 -0700
>
>
>
>
> User-agent: 
>
> Mozilla/5.0 (X11; Linux x86_64; rv:52.0) Gecko/20100101 Thunderbird/52.2.1
>
>
>
>
> Michael Albinus wrote:
> > Paul Eggert<address@hidden>  writes:
> >
> > > Tramp needs to support a new method make-temp-file that creates a file
> > > (or directory) atomically, as make-temp-file does locally now.
> > That means, make-temp-file shall be converted into a magic file name
> > operation. I'll do my best, but I don't know whether we could get an
> > implementation for all Tramp methods w/o a race condition.
> >
> > > Tramp also needs to support the excl flag of write-region (currently
> > > it ignores that flag).
> > This sounds trivial. And yes, it will help.
>
> Thanks for doing that. Because of that, I'm no longer seeing a race in 
> make-temp-file, even for Tramp files. That is, although my proposed patch is 
> still a performance win on local files, I don't see how it is a correctness win 
> any more. It's still worth installing for the performance reasons, though, so I 
> did that and I am marking this bug as done.
>
> > > I was planning to write this up as a bug report after the patch goes
> > > in.
> > Pls do.
>
> To some extent this is moot now, if I understand things correctly. That is, the 
> only reason to write this up now would be for performance reasons, not a race 
> condition.
>
> There is still a race involving destination directories, for both Tramp and 
> non-Tramp versions. I plan to take a look at that next.
>
>
> --- End Message ---