Re: [PATCH] Add checks for new and tmp directories to mx_is_maildir()

Alejandro Colomar via Mutt-dev <[email protected]> Fri, 6 Feb 2026 16:13:37 +0100

Newsgroups	gmane.mail.mutt.devel
Message-ID	<qyot6xmzbrgplr2slmb5k4ktg6f4l3zkz677euguz56alxmw4b@szyjhrcger7r>

Hi,

On Fri, Feb 06, 2026 at 03:39:22PM +0100, Gero Treuner wrote:
> Hi all,
> 
> On Fri, Feb 06, 2026 at 02:37:36PM +0100, Alejandro Colomar via Mutt-dev wrote:
> > > > However, I personally argue for code clarity except when optimization is
> > > > called for.  In this case, I don't think it's worth the trade off.  See just
> > > > below the function in mh.c, mx_is_mh(), for similar code, also called by
> > > > mx_get_magic().
> > > > 
> > > > As always, if others disagree please just say so.
> > > 
> > > personally I see it the same way, considering readability and simplicity
> > > a big priority; parsing the proposed version is (at least for me) much
> > > harder than the original version.
> > 
> > +1
> > 
> > Except that maybe with fstatat(2) (as Crystal said) it could be readable
> > --we'd have to see the code, though--.
> > 
> > readability >>> optimizations
> 
> Thank you and others for the feedback.
> 
> A quick research shows that fstatat(2) isn't a good option:
> 
> 1) Even on Linux it was introduced no earlier than with glibc 2.4, so
>    this makes a case for portability fiddling with automake etc.
> 
> 2) The directory needs to be opened with a file descriptor, which adds
>    resource handling.
>    I don't see consistency of the directory in the event that by
>    mounting etc. the file system changes as a benefit here. IMO it is
>    fine to fail for a test of a mail folder when something looks fishy.
>    (Otherwise the test could succeed for a place which isn't available
>    any more.)
> 
> But the same gain of readability could probably be achieved by other
> means. (Let's see when I'll come back to puzzling ...)

If we can't have fstatat(2), I honsetly prefer the inefficient version.
I'd be surprised to change my mind on that.


Cheers,
Alex

> 
> 
> Kind regards,
>    Gero

-- 
<https://www.alejandro-colomar.es>
Use port 80 (that is, <...:80/>).

signature.asc (application/pgp-signature, 833 B)