Accept this agenda?

-> http://www.w3.org/XML/XProc/2014/08/27-agenda

Accepted.

Accept minutes from the previous meeting?

-> http://www.w3.org/XML/XProc/2014/08/20-minutes

Accepted.

Next meeting: 3 Sep 2014

<ht> Regrets for 3 Sept

Norm gives probable regrets for 10, 17, and possibly 24 Sep

Specification issues

-> https://github.com/xproc/specification/issues

Norm describes the move to GitHub issues for the spec

Alex: Did you merge my XInclude stuff?

Norm: Yes, in the xproc20 branch.
... I want everyone to use pull requests to change the spec.

Test suite issues

-> https://github.com/xproc/test-suite/issues

Norm describes the test-suite repo

-> https://github.com/xproc/test-suite/pull/13

Norm describes his pull request.

-> https://github.com/ndw/test-suite/commit/018adf51bb43edea7c00883818507bea6ddbbf6b

Alex: Alternatively, we could have multiple expected results and the test passes if you match any one of them.

Norm: I agree, but that'd be a bigger change in the test suite.

Alex: I'm fine with this change, but we should consider multiple outputs for the future

-> https://github.com/xproc/test-suite/issues/15

Norm: I propose that we accept the change offered in https://github.com/xproc/test-suite/pull/13

http://tests.xproc.org/tests/required/data-002.xml

Alex: How far are we going to go, are we going to try to produce tests that test every conceivable variation.

Norm: We're going to go as far as we can.

Henry: I hear Norm saying "it would be a bug if our core processing didn't accept quoted content type parameters", but that doesn't appear to be the problem.
... Alex is suggesting we should normalize these things. If we find places where there was a bug in somebody's step implementation then we should try to make sure that we test that.
... But we can't go and look for all possible bugs.
... What I heard Alex say was that we should actually normalize this.

Alex: I think you misheard me. I think there is an open issue of how we deal with this in general in our test suite for 2.0.
... I think we should consider this carefully so that we have a good test suite without combinatorial issues.
... My question was "should we be trying to be that complete". Norm says "best effort" and I think that's a fine answer.

Norm: I'm happy to have tests for all the cases we think of.

Alex: I've run into this with other test suites.

Norm: I think experience with the 1.0 test suite will improve the 2.0 test suite.
... Coming back around to my question, does anyone object to the proposed change to those two tests?

None heard. Accepted.

-> Requirement “2.3 Align with XPath 3.0 technlogies”

-> https://github.com/xproc/specification/issues/30

Norm: I attempted to address part of the 2.3 requirement, removing XPath 1.0

-> https://ndw.github.io/specification/langspec/remove-xpath-10/head/

Norm waffles on a bit about XPath versions...

Alex: How are we going to track the changes? For example, if I wanted to write an upgrade tool, how would I know that two p: functions had been removed.

Norm: I guess there should be a changelog.

Alex: It doesn't have to be in the document.

Any other business?

None heard.

Adjourned.

rrsangent, draft minutes