Just to remind you that we need to gather proposals for new project names and any split in the functionality (i.e. separating schedules from service following).
If I dont receive any suggestions by this point I shall conclude that the current project name and structure is the most appropriate. Please feel free to contact me in private or in a public reply to the group. Public is better as this then gets discussion going.
As far as I understand RadioEPG arose from the idea of providing
"schedules". Additionally a solution was investigated how to provide
"service linking" with the same peace of technology. Currently it clearly
looks like the "service" part is much more important whilst the "schedules"
part already became optional.
From the reverse angle we're naming a "service following" specification
"RadioEPG" ... whereas schedules needn't be included. From my (and ARD's)
point of view that's not a good idea.
Providing RadioEPG with no schedules included would be very strange ... not
only from a marketing pov. Much more important for us is the fact, that
service linking/following is a too important part of radio and a crucial
requirement for the German market so that mixing it up with something else
is completely out of question.
That's why we're proposing to seperate those two parts
RadioINFO- information about the STATION is mandatory- information about its PROGRAMMES is optional
Many thanks for your honest answers ... but I can't acknowledge my defeat
so easily :-)
Ben, you're the RadioEPG team leader so I completely trust you if you say
that seperating those two would yield in large effort with little return.
But this would be a problem of the RadioDNS project ( ... and of your
person. I know - sorry ... I don't want to make your life miserable).
However - I still feel that seperating the specs would be easier to sell at
least to the broadcasters. Please help me out ... I read the spec a few
times but I'm not feeling certain at the moment: If I was a broadcaster
willing to reuse my existing dab epg pi/si files ... what would I exactly
and mandatorily have to change in my files if I'd waive the service linking
stuff and solely focus on the schedules?
James, I don't agree with you either because we have that requirement: I've
brought it in several times :-)
Seriously - I don't think that you (nor the RadioDNS project) should make
decisions for other markets e.g. whether programme information are useless
or not. We both agree that it's a marketing issue. Good ... epg and service
linking are two independant functionalities in dab (which is from my
understanding the only relevant radio broadcasting technology ... besides
fm that unfortunately can't carry pi in the broadcast channel). So I'm
simply proposing that we shouldn't mix both things within our project.
Making epg to a subset is a meaningful decision.
Please be assured that I fully support the goal of the RadioDNS project.
This is exactly what I want to do ... especially looking at the German
market which I'm quite familiar with. I would be keen to learn what would
make the implementation harder for the broadcasters if we'd seperate specs?
Currently I can only see that the implementation is harder when we have a
mutual spec ...
Best
Sebastian.
James, I don't agree with you either because we have that requirement: I've
brought it in several times :-)
Seriously - I don't think that you (nor the RadioDNS project) should make
decisions for other markets e.g. whether programme information are useless
or not.
We both agree that it's a marketing issue. Good ... epg and service
linking are two independant functionalities in dab
Many thanks for your answer which really helps me to understand your view.
Interestingly I don't think that a programme guide is an optional subset of
station information. I think that service information is just a necessary
piece of data to make use of the programme information. But you're
right ... that needn't be the case in the IP-world. However - from my
understanding the RadioDNS project's foundation is broadcasting.
So pushing everything together into one spec is pretty sensible when you're
on a "greenfield". But I'm arguing from a broadcasters position that
already has an evolved production and transmission infrastructure ... e.g.
distributes an epg (si and pi files) for its broadcast transmission. Such a
broadcaster wouldn't like to revise any of the already produced data when
it is so close to an already existing standard. With the current proposal
he has to ... which honestly creates some extra effort (and hopefully Ben
can give us some guidance on how much it really is).
What do the others think ...?
Best
Sebastian.
Interestingly I don't think that a programme guide is an optional subset of
station information. I think that service information is just a necessary
piece of data to make use of the programme information.
So pushing everything together into one spec is pretty sensible when you're
on a "greenfield". But I'm arguing from a broadcasters position that
already has an evolved production and transmission infrastructure ... e.g.
distributes an epg (si and pi files) for its broadcast transmission. Such a
broadcaster wouldn't like to revise any of the already produced data when
it is so close to an already existing standard. With the current proposal
he has to ... which honestly creates some extra effort (and hopefully Ben
can give us some guidance on how much it really is).
Probably best answered with a practical point of view.
Speaking with my broadcaster's hat on (it has a transmission tower on
it), we have managed to fit in RadioEPG into our process with minimal
effort, but the actual level of effort is dependant on how
closely-coupled your information source is to your output.
We (Global Radio) have a set of sources to provide us with Station
information, stuff about platforms and bearers, and schedules.
These we use to output to various places - websites, satellite EPG,
directory service providers (in their many and varied formats), and
DAB EPG. We added to this an output for RadioEPG.
The use of the same XML entities and structures allowed us to use a
common set of tools across DAB EPG and RadioEPG, especially in the
generation of PI files (it is open source too, which is nice). We used
the same tool we had for SI generation *as a basis for* our XSI
generation tool.
Basing the RadioEPG specification on DAB EPG and using common
entities, structures and concepts allowed us a lot of reuse both in
terms of the actual code in generating data and also the logic in the
way that things are done. We couldn't have done this if we would have
started with brand new specification, with entirely different
entities.
Even if I were a broadcaster with an entirely closed EPG chain, such
that all I have is a bunch of binary DAB EPG PI and SI files, it would
still be possible to take these, add the necessary information to them
and regenerate to the necessary XML files for RadioEPG. Again - there
are freely available tools to help you do this.
Ben
Just a slight important clarification, but one thats important to keep in mind:
*You **can** keep the exact same PI file as you do for DAB EPG*. This
is because if no bearers are specified in the PI file, bearers will be
inherited from the SI file in DAB EPG and the XSI file in RadioEPG.
The XSI file is very close to the SI file format - services are placed
inside a <services> element rather than an <ensemble> element and so a
simple XML transform can produce XSI from SI (or vice-versa).
Ben
Thanks for this good discussion so far ...
>> Ace. So we're agreed that SI is mandatory if we're to have a programme
schedule. We don't appear to agree that a programme schedule might be
optional; but I hope you understand my point of view here that for some
broadcasters, mandating a programme schedule makes things difficult (and
subjects them to unnecessary work).
Yes - absolutely. We should keep the hurdles as low as possible and avoid
unnecessary work. So we shouldn't force anybody to provide a programme
schedule if he or she doesn't want to. But that doesn't exclude my
proposal. I also don't want to force anybody to provide anything
unnecessary. Quite the opposite ... I'm proposing to let everybody use the
'old and untouched stuff' except he or she wants to do more ... because ...
>> Inevitably, both the SI and the programme schedule bit will continue to
evolve;
I'd also agree. However - do you really feel that the programme schedule
will continue to evolve ... given that it seems not so important for the
majority of us than the XSI and service linking bit? So following the
opinion of our IT department when it comes to hardware procurement I'd say:
"one function per device" ... i.e. "one function per spec" in our case.
Assuming that it'd make things less complicated as we can then evolve one
bit without necessarily adapting any changes within the other bit (which
would probably also require e.g. my company to do unnecessary adaptions).
>> *You **can** keep the exact same PI file as you do for DAB EPG*. This is
because if no bearers are specified in the PI file, bearers will be
inherited from the SI file in DAB EPG and the XSI file in RadioEPG.
Many thanks for that. So I understand we can reuse our existing dab-epg
pi-files without changing anything (with the restriction that the bearer
info on pi-level would be impossible then) and only would need to revise
our existing si-file to an xsi-file (which is the two-minute-work). I also
understand that the big effort in splitting the specs would reside in
reproducing/emulating the bearer info on pi-level. Is that right?! The
reproduction of the bearer info on global level (i.e. how it's currently
done with the xsi) would be a minor task out of my understanding. Is that
also correct?!
Best
Sebastian.
So following the
opinion of our IT department when it comes to hardware procurement I'd say:
"one function per device" ... i.e. "one function per spec" in our case.
Back to the naming ... I'm fine with RadioESG. The spec incorporates
'service linking' and 'epg' functionalities (whether optional or not). So
any name should reflect both. RadioESG seems to fit best ...
Best
Sebastian.