auto-update in 0.16.2

[David Relson]

On Sun, 18 Jan 2004 01:21:33 +0100
Stefan Bellon <sbellon at sbellon.de> wrote:

> Stefan Bellon wrote:
> > David Relson wrote:
> 
> > > Introducing a config file option "thresh_update=0.01" and a
> > > corresponding command line option "-u 0.01" seemed the obvious way
> > > of dealing with this. Cutting the number of wordlist updates has
> > > the dual benefits of making bogofilter faster and slowing the
> > > growth of the database.  I thought it a worthwhile addition :-)
> 
> > It is, it is, just not as optional argument to the switch -u.
> > Pity that -U was used for something else a few weeks ago.
> 
> > And apart from the used letter, I think the line in the usage output
> > should go into the "parameter options" section and not directly next
> > to -u.
> 
> I think I found another problem with the new thresh_update code. This
> time regarding the logging functionality. I just looked in my log file
> and noticed the following:
> 
> ...
> 18 Jan 00:54:17 006 X-Bogosity: Yes, spamicity=1.000000,
> version=0.16.3-sb1, 
> 18 Jan 00:54:18 006 X-Bogosity: Yes, spamicity=1.000000,
> version=0.16.3-sb1, 
> 18 Jan 00:54:19 006 X-Bogosity: Yes, spamicity=1.000000,
> version=0.16.3-sb1, 
> 18 Jan 00:54:19 006 X-Bogosity: Yes, spamicity=1.000000,
> version=0.16.3-sb1, 
> 18 Jan 00:54:21 006 X-Bogosity: Yes, spamicity=0.988504,
> version=0.16.3-sb1, register-s, 324 words, 1 messages
> 18 Jan 00:54:21 006 X-Bogosity: Yes, spamicity=1.000000,
> version=0.16.3-sb1, register-s, 324 words, 1 messages
> 18 Jan 00:54:23 006 X-Bogosity: No, spamicity=0.000000,
> version=0.16.3-sb1, register-s, 324 words, 1 messages
> 18 Jan 00:54:24 006 X-Bogosity: Yes, spamicity=1.000000,
> version=0.16.3-sb1, register-s, 324 words, 1 messages
> ...
> 
> Those have been processed in a batch with -pluM as command line
> options. It looks like from the point on when one message was
> registered, all others are logged as registered as well though they
> are not.
> 
> I think the problem may be in passthrough.c write_log_message() where
> old contents of the msg_register buffer are appended to the log
> message. Perhaps run_type is RUN_UPDATE even if thresh_update disables
> the update itself? What do you think?

Sounds very possible.  I've just used it in single message mode.

Try setting the first byte of each buffer to '\0' after writing the
message and let me know what happens.