makehistory currently outputs (many) broken lines.
Dan Merillat
harik at chaos.ao.net
Mon Jan 17 21:40:23 UTC 2000
I cvsuped the tree mid-last-night and compiled... makehistory is producing
extremely badly malformed lines. (around 20 over the 3million lines in history)
These lines kill makedbz, requireing manual intervention. Speaking of DBZ, is
it possible to have it spit out the tokens of duplicates when it encounters them?
I've patched it to do that before (back when it was still part of makehistory)
Examples are
"0000@" <- just the tail end of the token on it's own.
If I get time tonight, I'm going to look into the file-io of makehistory
to see if I can drop a format checker in there, to verify each line before
it hits the history file.
Also, in the spirit of allowing things to work, can makedbz ignore malformed
lines and just go to the next valid one? Obviously, a corrupted history file
is not desirable, but a running system without 2 hours of manually editing
is.
Running a linux/libc6 system and --with-perl
--Dan
More information about the inn-workers
mailing list