comprehensive testing

[Matthias Andree]

David Relson <relson at osagesoftware.com> writes:

> I've got a different 'comprehensive' set which tests tdb, qdbm, sqlite,
> db4.[123], with each db4 version tested for --enable- transactions, --
> disable-transactions, and AUTO-XA mode.
>
> My tests show a regression:
>
> ./configure --enable-transactions
> make -s check
> ...
> bogofilter: (db) DB->get(TXN=134690040,  '.MSG_COUNT' ), err: DB_PAGE_NOTFOUND: Requested page not found
> FAIL: t.bogoutil

I've seen this blink up once in a while, but never managed to get a hold
of it, and I'm not sure where it comes from. Does the problem persist if
you "make clean && ./config.status --recheck && ./config.status && make
-s" before "make -s check"? I'm not 100% confident the automake
dependency tracking catches all of our changes, particularly of headers,
and I regularly need to "make clean" on the Solaris machine to clear
spurious errors - during the current test round, Solaris failed to find
word_free() and word_dup() as it had not properly tracked the word.h
dependency down to transaction.c/.o and one other module.

Apparently, depcomp and automake aren't the surefire ways they create
the illusion of being...

-- 
Matthias Andree
_______________________________________________
Bogofilter-dev mailing list
Bogofilter-dev at bogofilter.org
http://www.bogofilter.org/mailman/listinfo/bogofilter-dev