Re: Request for clarification
Nick Morrott <[email protected]> Wed, 15 Apr 2020 03:51:43 +0100
| Newsgroups | gmane.comp.tv.xmltv.devel |
|---|---|
| Message-ID | <CAOQWjw3iJQccp31K_8a=bCeccNxSEJ3KfgVNRr8cJyy12K1e-g@mail.gmail.com> |
On Mon, 13 Apr 2020 at 19:12, Gary Buhrmaster <[email protected]> wrote: > > Before I submit a pull request fixing a grabber > that seems to be doing things wrong, I would > like to verify a few things IRT the dtd. @everyone If anyone feels the comments in the DTD are not clear enough, let's improve them. The DTD is nigh on 20 years old and TV has moved on significantly in terms of availability of platforms and channels within territories. > As I read the spec, a <previously-shown> tag should > *only* be used for repeats, and should never be used > for shows known to be first showings of the episode > (although the definition of first showings may be a bit > vague by itself in some cases). In the case I am > aware of, the previously-shown date is set to the > current date for these first showings (i.e. "we" are > the previously-shown). This seems to violate > the dtd statements and intent. My $0.10: if Marty McFly would call it a rerun, it uses <previously-shown>. Without looking, I'd guess most grabbers use the tag in its basic form to indicate a repeat, and do not set any start/channel attribute data. However, setting <previously-shown> to the <premiere> date makes no sense, falls foul of the DTD, and is incorrect. A known first showing (premiere?) can not have been shown previously. What constitutes a first showing is the $64,000 question. When determining what is considered a first showing (and thus the applicable scope of <previously-shown>), grabbers have (and potentially do) based this on: i) the territory the grabber serves (first showing on any channel in a country - the "UK premiere") ii) the network (first showing on broadcast TV - the "network premiere") iii) the channel (first showing on BBC One) History suggests it has been up to the data source and/or grabber maintainer(s) to pick one and use it consistently over time, if at all possible. When writing or updating a grabber, a maintainer likely sees how the data source provides its data, checks the DTD for available elements, and combines the two as they see fit. The problem is that data is not always perfect, the DTD is not always explicit, and the two may use identical terms for different purposes (e.g. "new"). The DTD comments that the absence of the <previously-shown> does not mean a showing is a premiere. I'm speculating here, but I would imagine that if the listings contained an episode marked as S01E01 and not flagged as a repeat, most consumers of the data would treat it as a <premiere>, and some additionally as <new>. > In addition, as I read the spec, the <new> tag > should only be used for the very first episode > of a series (not every episode that is not a > repeat, that is determined by previously-shown?). > In fact that is explicitly stated as such, but I > want to be sure that that is what is really > intended. Yes. <new> should only be used for the "pilot/first" episode of a brand new series the *first time* it is broadcast. The DTD defines the scope to mean first time in a country. The country-wide scope of this tag makes it more explicit than either <premiere> or <previously-shown> tags, so some understanding of how the source data marks content as "new" is important if this tag is to be applied per the DTD as it stands. As you have inferred, subsequent first-run episodes of the new series are not <new> but could still be marked with <premiere>. Definitely not <previously-shown> until they are reshown on the same/different channel. Does any of this make sense? Cheers, Nick