Deb package

David Relson relson at osagesoftware.com
Wed Jan 29 01:29:33 CET 2003


At 06:58 PM 1/28/03, Clint Adams wrote:

> > Looks like 2 failures due to misuse to stupid sizeof() assumptions by
> > programmer (me) and 2 due to problems with database locking.  I'll let our
> > locking expert (Matthias) handle those.
>
>arm problem no longer occurring, waiting for mipsel data.
>
>New alpha and ia64 problems:
>
>http://buildd.debian.org/fetch.php?&pkg=bogofilter&ver=0.10.1.2-2&arch=alpha&stamp=1043796779&file=log&as=raw
>http://buildd.debian.org/fetch.php?&pkg=bogofilter&ver=0.10.1.2-2&arch=ia64&stamp=1043796945&file=log&as=raw
>
>I remember mentioning lack of IEEE FP on alpha a while back.  Could this
>be relevant?

It could be.  Some of the tests expect _exact_ matches in spam scores (to 6 
decimal places).  Expecting 6 digit accuracy probably isn't necessare and 
may be too demanding.  When someone shows me that 6 digits _is_ too 
demanding, I'll change it.

Interesting log files ...

alpha failed 3 tests:
         t.robx - no obvious reason
         t.systest - no obvious reason
         t.grftst - "Floating point exception"

ia64 failed 1 test:
         t.robx - no obvious reason

For each of the failed tests, you could (should) run it from the command 
line and then send me a tarball of the test.yyyymmdd directory generated by 
the test.  (Feel free to leave out the *list.db files :-)

David





More information about the bogofilter-dev mailing list