See also: IRC log
SA: Location can be single pointer, range,
shadow location, compound (for unstructured)
... todo: figure out composites. one location is for one mapping. several location pointers if there is more than instances
... misuse of compound location a problem?
SA: are there other types of location?
CI: heuristic tests in doc where error not located exactly - can we describe?
SA: probably possible with compound location
CI: case can occur where it's not clear what location type should be used
JL: no problem if we can not always point to the exact problem
SA: ambiguity on the location types, clarify
which type to which scenario to use
... will diff developers use different loc types?
<niq> something like line/char or byte-offsets can't hope to correspond exactly over different tools
<niq> see the example in what I posted before: different validators
SA: <nik> ambiguity anyway with respect to line/char byte-offset, etc.
<niq> urls at the bottom of http://lists.w3.org/Archives/Public/public-wai-ert/2005Sep/0012.html
<niq> not big problem, it's the same error
<niq> v.w.o: "Line 10 column 20:"
<niq> valet: "Line 11 char 1"
SA: have to provide good description of how to use line etc.
<niq> htmlhelp: "Line 11 char 2"
CI: problem for e.g. one specific testcase - and not knowing how to handle
SA: tool would expect one of the four locs.
CI: will tools understand error independent of the location types used to describe it
SA: all agree
... earl consuming tools must be able to handle all different loc. types
... next steps, refining classes, properties, etc
<niq> hmmm ....
<niq> will think about it, not sure if I have much to add to earlier discussion
SA: working on everything that belongs to single loc. class?
<niq> ah, that's what you're talkin'bout
SA: e.g. line (domain, range, line col have superclass)
<niq> can give that a try ...
<scribe> ACTION: work on sinle location location [recorded in http://www.w3.org/2005/10/04-er-minutes.html#action01]
<niq> *and* 'cos chaals disagreed with everyone ...
SA: ruleset part of testcase - special part of testcase?
SA: waiting for charles' responds, continue on
later call or f2f
... overall structure of the evidence class: properties to ease queries
SA: what to do with annotated reports - add assertor?
<niq> audit trail for assertions/assertors?
<niq> it's a new assertion - the previous assertion is evidence for the new one, yesno?
JL: modified assertion - more sort of signing
the report (outside of the earl model)
... like signature of the file, trusted places, etc.
<niq> talking of trust, should we support pgp-signed reports?
SA: add evidence
<niq> aob is always the best item
<niq> beer brakes if you please - this is .de!
<JibberJim> Hmm, I wonder if I could get from CGN to the meeting on the monday morning...