Is non-zero entry point on subtitles a thing now?

[IoannisSyrogiannis]

First of all, I should write that I am aware of:
SMPTE's RDD 52 of 2020, paragraph 7.2.3 about StartTime Element reading that
The StartTime element shall be present, and the value shall be 00:00:00:ZeroE as defined In SMPTE ST 428-7
as well as:
SMPTE's RDD 52 of 2020, paragraph 8.3.2 about Entry Point reading that
For all MainSubtitle or ClosedCaption timed text tracks, the Composition Playlist’s EntryPoint element as defined in SMPTE ST 429-7 shall be present and have a value of “0”.
and:
Deluxe Recommended Guidlines for DC source and Content of 2022, paragraph 3.2.7.1 about CPL Entry Point reading that EntryPoint in a CPL referencing any Timed Text asset may never contain a non-zero value. This applies to both Subtitles and both Open or Closed Captions.

Contrary to those, though, it came to my knowledge that a relatively recent version of easyDCP was using on its CPL an entry point of one hour.

If you wonder, what the devil could have triggered it, I would suppose that this DCP or VF was made on DaVinci Resolve and the original had an offset of one hour, as many timelines have preset. As a result, the subtitles entered were synched with such an offset and -upon „delivery“- the entry point was used to compensate for that offset without eliminating it while making the DCP.
(After all, if that is indeed the case, the video and audio had to go through that offset elimination.)

That was my first surprise, that wouldn't have been so big if it was Clipster, or QubeMaster (I have limited experience with Transkoder), since it has happened before that someone, somewhere would have made a mistake (who doesn't) and a „broken“ DCP would have arrived to my doorstep. Yet, easyDCP, being so strict with recommendations to call anything straying off them „error“ and reject it, behaving casually about such a „shall“ from SMPTE standard 429-7, that seems even more extraordinary.

My second surprise, even more striking, was that a server belonging to the generation of those that still have regular updates (that is to say, not a Doremi, or a DSS played those subtitles correctly! It took note of that non-zero entry point and acted as asked, screening those subtitles on sync.

On all those, I am just an observer. I do not work with this specific DCP or cinema server in question. Yet, knowing about it, I can't help but wonder:
Is there some new rule, or a loosened restriction that made that „shall“ become „may“ or „can“, in some recent 429-7 edition?
Can anyone shed some light on the matter and its implications?

[Carsten]

I remember an ISDCF meeting where it was discussed that these constraints are in place because some prominent server needs it to be like this. That was a few years ago, and the server was probably a Doremi or DSS200. Don't know wether that behaviour has later been fixed with a software update. There are other timed-text constraints as well.