Re: Now playing: miltertest

From: Daniel Black <daniel.subs_at_internode.on.net>
Date: Thu, 3 Dec 2009 05:43:54 +1100

On Wednesday 02 December 2009 05:44:51 Murray S. Kucherawy wrote:
> On branch br-msk-lua I've whipped up a milter-level filter test facility
> that's Lua-based.

great work Murray.

> I intend to include a directory under "opendkim" called "tests" which has
> a comprehensive battery of unit tests done at the milter level that
> exercise the filter's features, something we're currently missing.

Sounds good. I'll get the test coverage code running there too (next week
sometime).

> Daniel had some ideas about doing full integration testing in an automated
> fashion. That's still probably a good idea, but it's harder to do that in
> a self-contained fashion given the complexity of a full-blown MTA.
True - I'll stage this in a different directory. I'll start with some test
emails, and a script to receive the email at the other end and save to a file.
and a bit of logic to compare the two.

libinterpose is something I came across at OSDC. It basically uses LD_PRELOADs
to load a malloc library wrapper that does a fork, with the parent returning
success and the child failing. By using a similar strategy for other system
calls we can get a much better test coverage. Even the current DNS answer
injection test code could be rewritten.

http://2009.osdc.com.au/clinton-roy

The other GLIBC tool to deal with memory allocation tracing was mtrace. With
this you compile as normal. Use a mtrace() function call at the program start.
set the env variable MALLOC_TRACE=filename.log and run the programs. addr2line
makes a pretty output.
Received on Wed Dec 02 2009 - 18:44:05 PST

This archive was generated by hypermail 2.3.0 : Mon Oct 29 2012 - 23:32:30 PST