[ET Trac] [Einstein Toolkit] #114: per-variable tolerances for Cactus testsuites

[Einstein Toolkit]

#114: per-variable tolerances for Cactus testsuites
--------------------------+-------------------------------------------------
  Reporter:  knarf        |       Owner:  tbode     
      Type:  enhancement  |      Status:  review    
  Priority:  major        |   Milestone:  ET_2011_11
 Component:  Cactus       |     Version:            
Resolution:               |    Keywords:            
--------------------------+-------------------------------------------------

Comment (by hinder):

 OK, in that case, the debugging code should be left in.  I didn't look too
 closely at that, it was just a general observation.

 I hadn't considered the case of files matching more than one expression.
 As you say, in that case probably the code should abort and the user
 should make the expressions more specific.  If this becomes a problem in
 future, we can relax the restriction and print a warning that only the
 last expression is being used.

 The reason I said that the code wouldn't work for multiple regexps is that
 I had misunderstood the code on line 219.  Sorry.  You are correct, it
 should work fine.

 I disagree that my suggestion would invalidate all current test files.  My
 proposed interface includes the original interface as a special case -
 i.e. the user specifies ABSTOL = 1e-9.  They can optionally put a regexp
 after it if they want to use the new feature.  I agree that the opposite
 order is a bit more logical, but that would mean that you couldn't
 interpret this as an additional optional positional argument which
 defaulted to '.*'.  I think that the new behaviour is just an extension of
 the old, so a new keyword is unnecessary.  It also means that you need
 separate code to handle the two cases, whereas if you have the same
 keyword, you can use the new code exclusively, with just one additional
 line to say "the default regexp is '.*'".  This reduces code duplication
 (you can then remove the current ABSTOL and RELTOL cases in the code).
 Unless I have missed something else!

-- 
Ticket URL: <https://trac.einsteintoolkit.org/ticket/114#comment:6>
Einstein Toolkit <http://einsteintoolkit.org>
The Einstein Toolkit