diff options
| -rw-r--r-- | NEWS | 50 | ||||
| -rw-r--r-- | fetchmail.man | 4 |
2 files changed, 44 insertions, 10 deletions
@@ -1,29 +1,63 @@ - To Do: + Complaints and Suspected Bugs: -George Sipe reports that --monitor is broken. +Reports from from Klee Dienes <klee@mit.edu> on the Debian bug list and from +fetchmail-friend Dirk Eddelbuettel <edd@rosebud.sps.queensu.ca>: -This needs to be fixed. --interface seems to work OK, so the error is -evidently something pretty specific to the --monitor code. +[KD1] If the local SMTP server disconnects while fetchmail is initializing +the connection (for example, because inetd answers the connection but +can't find the appropriate binary), fetchmail will wait forever for a +response from the server. + +(Hm. Possibly this is the same bug as DE2 below. It appears the server +timeout feature may be broken, but I have not been able to reproduce this yet. +It's hard to test without a customized server or a very overloaded or flaky +one. The one I use is too fast. :-)) + +[KD2] Also, fetchmail seems to be using a polling loop to wait for responses +from the server. This should probably be changed to a select()-based +mechanism. + +(Perhaps. This is a design change I'm not ready to do quickly.) + +[DE1] After starting + + fetchmail --syslog --monitor ppp0 --daemon 300 --timeout 30 + +I said 'fetchmail --quit' but that isn't logged. + +(I haven't been able to reproduce this. I ran `fetchmail -d 2 -N -v' +in one window and then did `fetchmail -q' in another, and the log +message "fetchmail: terminated with signal 15" did show up. What +happens when you try that?) + +[DE2] I had requested a timeout of 30 secs, but it didn't time out in almot + fifteen minutes + +(But fetchmail -V reports the timeout option set correctly. Are server +timeouts really broken? We must find a way to test this.) Release Notes: ------------------------------------------------------------------------------ -fetchmail-3.8 () +fetchmail-3.8 (Fri Mar 7 17:34:02 EST 1997) features -- * More FAQ material on using `localdomains'. -* compilation hacks for ISC 4.0 (thanks, Larry Jones!) +* compilation hacks for ISC 4.0 (thanks, Larry Jones!). bugs -- -* enabled ETRN and RPOP command-line option +* enabled ETRN and RPOP command-line options. * yet another attempt to fix the error.c compilation problems under Solaris and NEXTSTEP. -* handle \( and \) correctly in RFC822 comments. +* handle \( and \) correctly in RFC822 comments, thanks to Gareth McCaughan. + +* Fixed off-by-one error fingered by Brian Jones that prevented + `localdomains' from working. ------------------------------------------------------------------------------ fetchmail-3.7 (Fri Feb 21 17:38:40 EST 1997) diff --git a/fetchmail.man b/fetchmail.man index ebc68127..b1fb6014 100644 --- a/fetchmail.man +++ b/fetchmail.man @@ -247,8 +247,8 @@ under Linux. Daemon mode can cause transient links which are automatically taken down after a period of inactivity (e.g. PPP links) to remain up indefinitely. This option identifies a system TCP/IP interface to be -monitored for activity. After each poll interval, if the link is up and -no other activity has occurred on the link then the poll will be +monitored for activity. After each poll interval, if the link is up but +no other activity has occurred on the link, then the poll will be skipped. This option is currently only supported under Linux. .TP .B \-A, --auth |