RE: [TEC 3.9] Fewer large baroc files vs. many small baroc files?

The rb_dir thing is kind of embarrassing :(

I realized that just after I had sent the original note 
(kind of like slamming your car door and realizing the 
keys are still in the ignition just as your fingers 
release the door handle - oops). That's what I get for
posting as I was was putting on my jacket to go home
for the night..

I have occasionally wondered how TEC handles the baroc
files on startup, but never really ran any tests to 
determine how it actually worked. I had hoped that the
TEC_T_ISA table was only rebuilt when the rulebase was
changed/compiled/loaded, but I guess that's not the case.

Upon further reflection, I may rethink this whole idea
anyway. Even though my intent was to include the baroc
source in the "Information" files (similar to what I do
with rules and Prolog predicate documentation, which 
can also be accessed via the "Information" button), I'm
now thinking that a single class definition may not be
sufficient. To be truly useful, the "Information" page 
ought to show the entire parentage for the event, tracing
its lineage all the way back to the base EVENT class. I
think that will take some fancy cipherin' on my part to 
generate all these docs from the baroc files.

I shall ponder this further..

Loren Cain
Digicon

-----Original Message-----
From: owner-tme10-XtjxT7Vmt5b1ENwx4SLHqw@xxxxxxxxxxxxxxxx 
[mailto:owner-tme10-XtjxT7Vmt5b1ENwx4SLHqw@xxxxxxxxxxxxxxxx]
On Behalf Of Thomas Seeling
Sent: Tuesday, March 22, 2005 1:49 AM
To: tme10-XtjxT7Vmt5b1ENwx4SLHqw@xxxxxxxxxxxxxxxx
Subject: [tme10] [TEC 3.9] Fewer large baroc files vs. many small baroc
files?

Hallo,


>restructuring my baroc files so that I have a separate file for each 
event class
>in fact, after compile/load, you can delete the entire TEC_CLASSES 
directory(s)
> and TEC does not seem to notice until the next time you compile/load).

yes, but try to delete $DBDIR/tec/rb_dir/TEC_CLASSES and be surprised
when 
restarting ;)

TEC parses the files on startup and inserts all classes in TEC_T_ISA
with their parent/child relationship, so the files are definitely 
required.
I think your approach would work, but imagine 5000-15000 files in a 
directory.
This is quite some overhead for the OS, and imagine the upgrade fun
when *any* product changes its classes. You'd have to identify which
class has changed, or simply decompose the files from the upgrade
and reapply your changes -- imho a nightmare.

I'd vote for keeping the files as-is (as-are? ;) ) and decompose using a

BNF parser
when you need it.

Nice idea to include baroc source with the "information" button ;).

Tschau...Thomas
-- 
"Do you wanna be a legend or a passing footprint on the sands of time?"

Senior Consultant, Tivoli Certified Enterprise Consultant + Instructor
santix AG, www.santix.de, info-/JKUMw0Y9jazQB+pC5nmwQ@xxxxxxxxxxxxxxxx, fon 
+49-89-321506-0, fax -99
Weihenstephaner Str. 4, D-85716 Unterschleissheim, GSM +49-170-9135811