Topic and group parameters

[rintaman@cs.Helsinki.FI (Janne Rinta-Mänty)]

Hello,

The order in which the value of a parameter is determined seems to be
(strongest first):

1. group parameters
2. topic parameters (from deepest to root topic)
3. gnus-parameters (in the order of the list)

...but is it?  Experiments with gcc-self (the columns are the values
of gcc-self in group parameters, topic parameters and gnus-parameters
('-' means no value at all); the last column is the value seen in Gcc:
when writing an article in the group (nil means no Gcc header)):

group     topic     g-p     Gcc

"group"   "topic"   "g-p"   "group"
"group"   nil       "g-p"   "group"
"group"   -         "g-p"   "group"
"group"   "topic"   -       "group"
"group"   "topic"   nil     "group"
"group"   nil       nil     "group"
-         "topic"   "g-p"   "topic"
-         "topic"   nil     "topic"
nil       "topic"   nil     nil
nil       "topic"   -       nil
-         "topic"   -       "topic"
nil       "topic"   "g-p"   "g-p" ???
-         nil       "g-p"   "g-p" ???
nil       nil       "g-p"   "g-p" ???

Last three values for Gcc look odd, I was expecting no Gcc header for
those combinations.

Also, I think that gnus-parameters should be stronger than topic
parameters.  The manual says in <info://Gnus/Topic+Parameters> "Group
parameters (of course) override topic parameters [...]", and
gnus-parameters has group parameters, only in a different place than
the "real" group parameters.

-- 
Janne Rinta-Mänty