bogofilter over NFS

[Matthias Andree]

Piotr KUCHARSKI <chopin at sgh.waw.pl> writes:

> We could try to first lock database, then open it and when this
> fails [mmap() will fail if file is fcntl()ed in scenarios where
> fcntl() fails because of file being mmap()ed], try db_open again
> without mmap. In this case we know fcntl() is alone and EAGAIN
> means someone's else lock and we can loop over fcntl() quite safely,
> while having possibility to reopen database, if it fails with
> mmap due to being fcntl() locked.

The problem with this is that we may have nothing to lock before we
create the data base with db_open. Just opening and closing it will
choke libdb later, because it thinks it's being launched on a proper
data base, but doesn't find the pages it needs when the file is 0 size.
:-/

> I saw already. And really don't quite know why it hasn't been
> possible to do valgrind >vg.out itself

Because I only want valgrind output in there, not bogofilter command
output or stderr or something.

>> BTW, you cannot have valgrind because that's a tool that only works on
>> ix86 based Linux currently. You may have heard of the commercial Purify
>> tool; valgrind has similar goals.
>
> Purify, sure. Why in make check, then? Hm.

Regression tests...

-- 
Matthias Andree