I've been working with Richard on this problem of his,

Sunday, November 20, 2005, 11:09:40 AM, he ([EMAIL PROTECTED])
wrote: 

O> i'm building SA r345765 on OSX 10.4.3 w/ gcc401 & perl 587.
O> on '-lint', i get the following previously unseen warnings:
O> ...
O> [20222] warn: Found = in conditional, should be == at (eval 3883) line 1046.

It only happens on SARE rules, and only on specific SARE files.
Nobody within SARE can reproduce it.

Richard was able to boil the problem down to:

> after some iterative hunt-n-pecking thru SARE_OBFU, i find that if
he runs --lint against:
> meta      SARE_OBFU_OBLIGATION     0
> meta      SARE_SUB_AGING_OB        0
> meta      SARE_SUB_FOR_OB          0
> meta      SARE_OBFU_BETTER_SUB     0
> meta      SARE_OBFU_CARTRDGE_SUB   0
> meta      SARE_OBFU_SEX_SPL        0
> meta      SARE_OBFU_TBL_05         0
> meta      SARE_OBFU_AMP            0
> and NOTHING MORE, i get the warnings (one per line above ...)

This is a structure I adopted a few months ago for the purpose of
removing rules without causing --lint errors when people had their own
score overrides or meta rules referring to rules being deleted.

The initial problem:  If I remove a rule for which there's a score
override, then --lint reports an unmatched score line, and RDJ will
not load any updates for that system.

The initial fix:  As discussed (I forget whether here or on the users'
list), the above "meta rulename 0" structure was adopted to say "this
rule exists, it always fails to match."

This structure has worked well, until now.  It fails as far as I know
only on Richard's
O> i'm building SA r345765 on OSX 10.4.3 w/ gcc401 & perl 587.

I cannot see any reason for these to fail on his system.  Can anyone
else find an explanation?

Is this a structure that should work?  If so, I'll create a bugzilla
entry for it.  If not, then I'll find an alternative.

Thanks.

Bob Menschel



Reply via email to