Re: [Evolution-hackers] Interface to Junk versus Not Junk buttons

[Joe Kelsey <joe zircon seattle wa us>]

On Mon, 2004-11-29 at 21:19 -0500, Daniel Gryniewicz wrote:
> On Mon, 2004-11-29 at 18:10 -0800, Joe Kelsey wrote:
> 
> > All it takes is an explanation in the documentation of exactly where and
> > when the system calls which external program with the exact arguments
> > used.  Once that documentation appears, then someone else can come along
> > and write something to perform a similar function.  You seem to feel
> > that doing something secretly is better than actually documenting what
> > you do.
> 
> man spamc

>From my reading of the source file, Evolution actually emulates a spamc
rather than actually calling spamc.  In other words, Evolution talks
directly to spamd.

However, that is not the real problem.  The real problem is how does
evolution deal with sa-learn.  There is absolutely no documentation or
any discussion on this mailing list that I can find which discusses the
actual interface to sa-learn that they chose to implement.

By spending hours looking through the mess of source code, I finally
stumbled across the actual part of the program which invokes sa-learn.
Not only did you chose the absolutely worst designed program with the
absolutely worst designed (or rather not designed) interface to use for
evolution, you did it without documenting your own usage, forcing anyone
who wanted to use a better designed program to hunt through undocumented
source code looking for the bits that they need.  A very simple one or
two paragraphs with the actual code listing showing how and why sa-learn
is invoked with different arguments would allow anyone to insert their
own interface.

As it stands, it looks like what someone has to do is reimplement spamd
since evolution pretends to be spamc.  Then you have to figure out the
idiotic sa-learn interface.  Then you have to hope that whoever is in
charge doesn't decide to fix this mess with a real design.

/Joe