W3C home > Mailing lists > Public > public-tt@w3.org > December 2017

Re: [ACTION 510] Use of TTML Version in Timed Text Toolkit (TTT)

From: Cyril Concolato <cconcolato@netflix.com>
Date: Fri, 1 Dec 2017 10:14:58 -0800
Message-ID: <CAMiyXwBTXCeamj_RrWk0PfWotgm7N09ox-zynH8fs9eWBsdYUg@mail.gmail.com>
To: Glenn Adams <glenn@skynav.com>
Cc: TTWG <public-tt@w3.org>
Hi Glenn,

Thank you for the detailed analysis. Some comments/questions below.

On Thu, Nov 30, 2017 at 6:14 AM, Glenn Adams <glenn@skynav.com> wrote:

> I have completed a review of the TTT codebase and find that the following
> uses of TTML specification version (as a applied to a specific document)
> are used during processing:
>    - [ttv] select between version specific default profiles
> I'm assuming default processor profile here as there is no default content
profile nor version-dependent algorithm in TTML2 to determine content
profiles, right?

>    - [ttv] select between TTML schemas associated with different
>    specification versions (56 instances)
> Initially, I thought that validation should rather be based on the
effective content profile and its related schema, but looking at the spec I
see that this effective content profile can be null. Is that one of the
issues that version is trying to solve? However, there does not seem to be
any normative algorithm to select the schema specified in the spec. Maybe
that should be added?

>    - [ttv] contextualize length expressions in terms of version (in order
>    to support new units) (2 instances)
> Do you mean that if you encounter a length value with rw or rh or without
unit in a document that has no profile and no version attribute, you treat
it as an error? Shouldn't this be done based on the presence of the
associated required feature in the effective processor profile, such as new
#length-integer-no-unit or #length-root-unit?

>    - [ttv] enable deprecation warnings (1 instance)
> I'm assuming this is related to <time-expression>, right?

>    - [ttpe] select use of initial value of tts:position versus tts:origin
>    (1 instance)
> As discussed in a previous call, shouldn't this be done based on the
support for #position, instead?

>    - [ttx] select between tt element and region element for root of style
>    inheritance (1 instance)
> Two remarks here:
- Couldn't this be done based on effective processor profile? If the
effective processor profile is based on TTML2 or above, select tt otherwise
select region?
- In any case, I could not find the text in TTML2 spec that says that if
there is no profile and the version is 1, tt is ignored for the style
inheritance. This should be added.

>    - [ttx] select between version specific transformation helper class
>    methods used during ISD transformation processing (24 instances)
> Could you elaborate on this point? I'm trying to understand if the TTML2
spec needs normative wording regarding this part or if it's an
implementation choice.

Received on Friday, 1 December 2017 18:15:21 UTC

This archive was generated by hypermail 2.3.1 : Friday, 1 December 2017 18:15:21 UTC