On Sat, 2008-09-13 at 16:25 +0800, [EMAIL PROTECTED] wrote:
> $ man bts
> DEBFULLNAME
> If DEBEMAIL is set, DEBFULLNAME is examined to determine the full
> name to use; if this is not set, bts attempts to determine a name
> from your passwd entry.
>
> "Examined": do mention that there are also angle brackets checks, etc.,
> so DEBFULLNAME in
> $ [EMAIL PROTECTED] DEBFULLNAME=Nurdson bts ...
> won't change anything.
In what way doesn't it change anything?
$ [EMAIL PROTECTED] DEBFULLNAME=Nurdson bts -n ret 12345 a
From: Nurdson <[EMAIL PROTECTED]>
> Also add an option to print everything that will be sent to the mailer
> so one can test. Or enhance -n, which doesn't show From: etc. currently.
-n *does* show everything that is sent to the mailer. If you don't have
DEB{EMAIL,FULLNAME} set then bts doesn't have any way of knowing what
address your MTA will use, so there is no From header in the generated
mail and therefore nothing to display.
> To do testing, one might think --sendmail=/bin/cat would work but it
> doesn't. Try it. Also --sendmail /bin/cat also didn't cause any
> errors. In both cases mail was sent to the BTS forthwith anyway.
It does work, so long as one has DEBEMAIL or DEBFULLNAME set. If they're
not then bts simply calls mail(1) directly rather than invoking sendmail
itself (and therefore doesn't use the command specified in --sendmail).
A quick scan of the changelog doesn't reveal any obvious reasons why we
don't just call sendmail in both cases (other than an old comment
referring to #18564, which the BTS doesn't have a record of). It appears
to be fairly easy to reuse the code path so --sendmail works in both
cases, but I need to do some more testing (and give my co-maintainers
chance to point out good reasons for keeping the split :-) before
committing the changes.
Regards,
Adam
--
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
