Duncan Findlay wrote:

[...]
> 
> The only thing is that we tend to have new features ready for release much
> faster, rather than waiting for hundreds of them, so this would be a
> problem, new features that are quite stable don't get to the users fast
> enough.
> 
> 


   Uh, based on Justin's new traveling schedule, and the crunch that 
practically everyone else on the devlopment team as said they've found 
them selves in, I suspect that the pace of releases going "golden" will 
slow down rapidly.
   Having a two-pronged development model allows us to get 
patches/fixes/increamental feature-adds quickly into production with a 
minimum of fuss/reconfiguration.  When you're working with a server(s) 
passing 10K-100K (1M?) messages a day, that's VERY important.
   I'd say that if you have "2.0.x" and you download "2.0.y" (where x < 
y) you should be able to do the "....; make install" and make little or 
no modifications to your configuration files to have a running system.
   If you go from "2.1.x" to "2.1.y" (where x < y) then there is 
potential for major changes, include restructuring/adding/dropping 
database tables, splitting up/merging/relocating cf files, 
upgrading/installing Perl or Perl modules.  The same holds true for 
going from "2.x.?" to "2.y.?" (where x < y).
   If these rules are followed, it will make SA much easier to 
administer and still allow rapid bug-fixes and feature adds (like the 
auto-whitelist and sql stuff) that are _very_ important to certain 
implimentations.
   Don





-- 
--------------------------------------------------------
Donald L. Greer, Jr                  [EMAIL PROTECTED]
System Administrator                 Voice: 512-300-0176
AustinTX                        http://www.AustinTX.COM/
   All opinions are my own.  Flame me directly.

"I don't necessarily believe software should be free...
but if you pay for it, it should work!" -- Me


_______________________________________________
Spamassassin-talk mailing list
[EMAIL PROTECTED]
https://lists.sourceforge.net/lists/listinfo/spamassassin-talk

Reply via email to