[I've sent this message twice, but I still don't see it in gmane or
 google groups...]

michael@cadilhac.name (Michaël Cadilhac) writes:

> When using `imap-shell-program' (with a nnimap-stream of 'shell), the
> process is created `as is' with start-process.  The docstring of
> start-process says :
>
>         If you want to separate standard output from standard error,
>         invoke the command through a shell and redirect one of them
>         using the shell syntax.
>
> Well, this is fun, because my distant imapd says the following :
>
> $ imapd
> * PREAUTH Ready.
> INFO: LOGIN, user=cadilhac, ip=[127.0.0.1], protocol=IMAP
>
> Oh, heck.  Gnus is searching for PREAUTH on the INFO line, doesn't found
> it, and just wait until the end of time to see what's happening next[1].
>
> Well, I'm not comfortable with the use of `2> /dev/null' on a package
> meant to work on Windows, so I propose this more radical change :

I've had a glimpse at Eshell, to see how it handles the 2> construction,
and... well... it doesn't.  So, I guess we can't do much better than the
proposed doc change.  Anyone against the commit?

(I'll install it in a few days if nobody complains.)

-- 
 |   Michaël `Micha' Cadilhac       |  La culture, c'est comme l'amour.      |
 |   http://michael.cadilhac.name   |     Il faut y aller à petits coups au  |
 |   JID/MSN:                       |  début pour bien en jouir plus tard.   |
 `----  michael.cadilhac@gmail.com  |          -- Pierre Desproges      -  --'