Title: Re: Summer of Code
Because the so many rules are tied to actual perl code via the evals, we've determined that the separation of .cf files as 'rules' and .pm files as code is arbitrary. Now there is a lot of support code in there that is obviously not rule related.
I suggest that your 'rule' project/release include any .pm files that end up being called from the .cf files
This would allow for more flexibility in changing the rules. Right now any rules that call functions in EvalTest are unchangable until there is a firmware update.
New rules that are better implemented as EvalTests are not includable until a firmware update.
There maybe some exceptions to this, possibly the bayes stuff since it involves all the database stuff, however for the most part, the eval rules do not depend on a lot of spamassassin firmware.
 
just a thought.
 
dale


From: Theo Van Dinter [mailto:[EMAIL PROTECTED]
Sent: Thu 6/2/2005 10:29 AM
To: [email protected]
Subject: Re: Summer of Code

On Thu, Jun 02, 2005 at 09:30:52AM -0700, Justin Mason wrote:
> can anyone think of good SpamAssassin projects for this?
> I'm a bit stumped ;)

There's a bunch of stuff we want to get done, I just don't know if any of it
really works well as a contained project.  A quick random brainstorm:

- Setup the backend of the updates system, get it working, start publishing
  updates for 3.0 and 3.1.
- Get short-circuiting working
- Part of me wants to split off the rules into a subproject, possibly doing
  releases via the updates system, but at least being able to move faster, do
  faster release cycles, etc.  More to the goal of "SpamAssassin" as engine as
  well.

An actual possibility is:

- Build a real/full test suite.  "make test" is a high level check for most
  things, but we've been talking about having a "several thousand" large test
  system to get down and check the nity-grity bits in the code.  Regressions
  for all fixed bug tickets, etc.

--
Randomly Generated Tagline:
 Earth Army Recruiting Center: What are you, chicken? Buk buk buk!

Reply via email to