> Thanks for the hint. Which reminds me: Is there some 
> kind of flowchart indicating which filters are being 
> applied first and how the software reacts if it 

The test routine consists of 5 different phases:

PHASE 1 : 
- expert "html only/non html only" option -> "i don't 
know" or "DO NOT catch" (abandon immediately)
- too far level reached -> same as above

PHASE 2 :
- check the link position (up/down/same address.. 
parameters ) -> "may catch" or "i dont know" status

PHASE 3 :
- "get non html files near a link" -> "may catch" 
or "i dont know" status

PHASE 4 :
- filters, external depth and robots.txt if used -
> "DO catch", "DO NOT catch" or "I don't know"

At this point, the engine makes its choice:
* if status == "may catch" ==> "DO catch"
* if status == "DO catch"  ==> "DO catch"
* if status == "DO NOT catch" ==> "DO NOT catch"
* else
if quiete wizard : "DO NOT catch" by default
wizard mode (asks questions) : asks the user!

PHASE 5 :
- if status is "DO NOT catch", and non html file 
encountered, and option "test all links", then ==> "DO 
test" (test mode)