hey,

For what it's worth, I am still +1 on the 'tvd-evaltoplugin' approach.

(I'm +0 on whether it should be 'all in one big LegacyEvalTestsPlugin.pm'
or 'split into multiple little .pms', though; I don't really mind either
way.)

By the way, looking back through that thread
(<http://marc.theaimsgroup.com/?l=spamassassin-devel&m=113581166500095&w=2>),
it's worth noting that some users of SpamAssassin *would* find it useful
to be able to turn off the existing eval test set: namely people who are
using the SpamAssassin engine, without using the basic SpamAssassin
ruleset.  For them, EvalTests.pm is all dead code.

And finally -- regardless of how the 'tvd-evaltoplugin' approach is
judged, the approach of putting *new* eval tests into plugin .pms can be
considered a quite separate issue, in my opinion.

--j.

Theo Van Dinter writes:
> On Thu, Jun 29, 2006 at 10:57:18AM +0100, Justin Mason wrote:
> > - (1) it breaks people running "sa-update" on an svn trunk build, since
> >   now the svn trunk update set includes references to an EvalTests.pm
> >   function that won't exist in older revs.   
> 
> Irrelevent IMO.  If one is running a dev version, there will be issues
> like this that have to be accepted.  sa-update is meant to work for
> released versions, not dev versions.
> 
> > - (2) I think we wanted to reduce the ongoing use of EvalTests.pm in
> >   general, as a goal, iirc.
> 
> That depends who you ask apparently (more below).
> 
> On Thu, Jun 29, 2006 at 04:21:54AM -0700, Loren Wilton wrote:
> > How hard would it be to make an evaltests plugin that literally contains
> > whatever rules are currently in evaltests.pm?
> 
> So ...  Last year at ApacheCon US a bunch of us had a discussion about this,
> and generally decided it would be best to get rid of EvalTests and move the
> functionality into plugins.  So I then spent 2 or 3 days worth of time going
> through and doing that, ending up with:
> 
> http://svn.apache.org/repos/asf/spamassassin/branches/tvd-evaltoplugin/
> 
> (now horribly out of date, btw.)
> 
> Everything looked good, except that certain people started complaining about
> the idea, the implementation, and generally were against getting rid of
> EvalTests.  It didn't really move forward from there, so that was the end of
> that.
> 
> > Yes, this is a kludge.  But if it can be done in 10 minutes vs 10 months or
> > even 10 days to split evaltests into 150 silly separate plugins, one for
> > each rule, then it might be the right thing to do.  It could be broken up
> > more later once it was out of the main project body.
> 
> I thought a plugin per rule would be really annoying, so I split it up
> into (iirc) 5 or 6 loosely related plugins (whitelist/blacklist, header
> rules, body rules, etc.)
> 
> On Thu, Jun 29, 2006 at 12:26:01PM +0100, Justin Mason wrote:
> > well, I'm not suggesting that we kill off the *current* contents of
> > EvalTests.  instead we can leave those there -- but future additions
> > should be kept out of that file if possible.
> 
> I'm +1 to killing off EvalTests, btw.  I argued about this before,
> so I don't want to fully rehash it again -- see the thread:
> 
> http://www.nabble.com/forum/ViewPost.jtp?post=2124116
> 
> I'd be willing to expend some effort doing this process again (only if we all
> agree (again?) ahead of time to do it).  The main time suck now is to merge
> the changes in trunk since the branch, which will likely be a few hours worth
> of work.
> 
> > As to the ease of moving the current contents into a new plugin .pm.
> > Actually this wouldn't be so easy, since people using a pre-move
> > copy of EvalTests.pm would then wind up with duplicated code; one
> > function in EvalTests.pm, and another one colliding with it from
> > the plugin .pm.
> 
> Again, irrelevent IMO.  If people are using the development version,
> they need to accept that sometimes things will blow up and that they will
> need to keep up with changes.  It's a development version for a reason.
> 
> -- 
> Randomly Generated Tagline:
> "The cardinal rule at our school is simple. No shooting at teachers. If
>  you have to shoot a gun, shoot it at a student or an administrator."
>                  - "Word Smart II", from Princeton Review Pub.

Reply via email to