[cups.bugs] [MOD] STR #1880: snmp backend fails to time itself out

[twaugh.redhat]

[STR New]

The try_connect() function in backend/snmp.c uses signal to set the alarm
handler.  However, even when this triggers, the connect() call is
automatically restarted -- and so does not get timed out.

I replaced the signal() call with sigaction() (and made sure not to use
the SA_RESTART flag) and this fixed the problem.

Original bug report:

  https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=200205

I also made a change to call try_connect() before httpConnect(), so that
we know the port is connectable before connecting.

Link: http://www.cups.org/str.php?L1880
Version: 1.2.2
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: cups-try_connect.patch
URL: <http://lists.cups.org/pipermail/cups-devel/attachments/20060803/3e8c0fe1/attachment.ksh>