- From: Thierry MICHEL <tmichel@w3.org>
- Date: Thu, 12 Mar 2015 10:57:39 +0100
- To: Nigel Megitt <nigel.megitt@bbc.co.uk>
- CC: Pierre-Anthony Lemieux <pal@sandflow.com>, Public TTWG List <public-tt@w3.org>
On 12/03/2015 10:48, Nigel Megitt wrote: > Note that the implementation tests can only verify that processors behave as specified when there is actually a specified behaviour. The "shall not"s you list actually state document features that implementations will logically never have to handle so there's no test needed. I agree that there is no behavior mentioned in the spec for these "shall not", therefore it will be difficult to show interoperatibilty of a particular behavior. In other Wgs I have seen tests of SHALL NOT assertions. Hence my question. > Unless you want to create a set of tests for document validation "transformation processors" which should fail documents that use the prohibited features? If it is only a matter of validation, then I guess we are fine. > > If you're going to list references for all the normative processor behaviour statements in the spec then at least all the sections I included in my analysis need to be present in the list. Those are not simply the parts of the spec that mention TTML feature or extension designators. We should demonstrate interoperability only for the extensions. > Nigel > > >> On 12 Mar 2015, at 09:36, Thierry MICHEL <tmichel@w3.org> wrote: >> >> Nigel, Pierre, >> >> Thanks for your report on the coverage of the Test suite vs IMSC1 CR. >> >> I have added in the implementation report [1] the following wording >> >> ["MSC1 introduces extensions to TTML1, as well as incorporates extensions specified in ST2052-1 and EBU-TT-D. Tests cover every feature defined in IMSC1 not already present in TTML1. Refer to the list of extensions added in IMSC1 in section 6.7, section 6.10, section 7.4, section 8.4 and in Appendix F. [http://www.w3.org/TR/ttml-imsc1/#features-and-extensions"] >> >> >> I have also added links in the table, to the appropriate section of the IMSC spec, for convinience. >> >> For this test suite we are mainly covering the SHALL assertions >> >> like for example: >> >> - The root container of a Document Instance SHALL be mapped to the related video object frame according to the following: >> - If ittp:aspectRatio is present, the root container SHALL be mapped to a rectangular area within the related video object such that: >> >> >> Shouldn't we also test the SHALL NOT assertions to test/describe the behavior ? >> >> For example: >> >> - Text Profile Constraints >> #image SHALL NOT be used. >> >> - Image Profile Constraints >> smpte:backgroundImageHorizontal and smpte:backgroundImageVertical SHALL NOT be used. >> smpte:image SHALL NOT be used. >> >> >> ittp:aspectRatio SHALL NOT be present if tts:extent is present. >> >> >> [1] https://www.w3.org/wiki/TimedText/IMSC1_Implementation_Report#Introduction >> Thierry. >> >> >> >> >> >> >> >>> On 12/03/2015 02:13, Pierre-Anthony Lemieux wrote: >>> Hi Nigel, >>> >>> Thanks for the review. >>> >>> I have added a test at [1] as recommended in row #1 of the report. >>> >>> I propose we discuss the other two test during our next call. >>> >>> Best, >>> >>> -- Pierre >>> >>> [1] https://www.w3.org/wiki/TimedText/IMSC1_Implementation_Report#Tests >>> >>>> On Wed, Mar 11, 2015 at 9:33 AM, Nigel Megitt <nigel.megitt@bbc.co.uk> wrote: >>>> All, >>>> >>>> Apologies for only being able to schedule this work so close to the >>>> deadline, but I've at last managed to do a pass through IMSC 1 looking for >>>> normative statements that affect processors, and reconcile this against >>>> the test suite that we have, as discussed in telcons recently. >>>> >>>> I think I've found some normative statements on processor behaviour for >>>> which tests are absent or incomplete. I attach a table showing this >>>> analysis. Just a reminder: the statements or features that need to be >>>> tested are those that are new to IMSC 1 relative to TTML1SE. >>>> >>>> In the case of the requirements from ยง6.6 concerning ttp:frameRate and >>>> ttp:frameRateMultiplier, I am unsure what the test should be because it is >>>> not clear exactly what the spec is requiring the processor to do: is it >>>> saying that the related video object must be advanced at the stated >>>> ttp:frameRate regardless of how it was encoded or decoded or is it saying >>>> that if they do not match already then the processor should display >>>> nothing, for example? >>>> >>>> Kind regards, >>>> >>>> Nigel >>>> >>>> -- >>>> Nigel Megitt >>>> Lead Technologist, BBC Technology >>>> Telephone: +44 (0)3030807996 >>>> BC4 A3 Broadcast Centre, Media Village, 201 Wood Lane, London W12 7TP >>>> >>>> >>>
Received on Thursday, 12 March 2015 09:57:42 UTC