Sean B. Palmer
2006-04-04 12:49:03 UTC
Hi,
On 18th January 2005, Christopher Schmidt and I, with the kind support
of Cody Woodard, released [1] a draft of RSS 1.1 [2] to this mailing
list. Every few weeks or so, either Chris or a member of the public
enquires, "what's the future for RSS 1.1? Why has it stopped being
developed?" (e.g. [3]), which I'm going to answer here.
RSS 1.1 was born of the fact of us liking RSS 1.0 but recognising that
it had many bugs, and a poor specification. I know many of the authors
of 1.0, and some of them are close personal friends; indeed, I learned
much of what I know about writing specifications from Aaron Swartz. I
have tremendous respect for them as writers of specifications, so I'm
not fully sure how RSS 1.0 came off so badly. Some of it was technical
limitation: I had recourse, for example, to RELAX NG Compact, which
the RSS WG did not. As for the rest, I'm too scared to look through
the archives to find out.
So when Chris and I expressed our mutual frustrations with 1.0 at one
another, we decided that rather than lauch yet another stonking
diatribe against the quality of the RSS 1.0 spec, as Bijan Parsia once
memorably and anyway unbeatably did, we thought we'd avoid the route
of the unmarried marriage counsellor and simply write a new
specification ourselves. And as a specification, 1.1 turned out to be
quite passable. Look kids, it has a conformance section!
We didn't know what to expect when we released the draft, but we had
an impression that we'd get the usual range of good, middling, and
ugly responses. In fact, we got far more ugly responses than I'd
expected--some quite considerably incompetent stuff to the point where
I gave up and let Chris deal with filtering it all.
On the good side, we got more implementations than expected, and a few
people embracing the new version. But the best of the feedback came
from the middling responses--the people who commented merely on the
technical issues of the specification and didn't bother to comment on
the politics of it. All of the bad responses that we got were either
political, or based on misfounded technical assumptions and ideas. It
all made me rather glad that I speak to people like Bjoern Hoehrmann
and John Cowan every day.
Chris and I wrote the specification and the validator and test suite
and all the implementations in about three days. The entry barrier for
producing a syndication language is not high, because getting it to
just work is not rocket science. That's why there are sundry versions
of RSS. Getting it to work perfectly, on the other hand, is a
Will-o'-the-wisp, and one which the Atom folk may be forever chasing,
though they're certainly the closest to it.
Anyway, the future. As I said to Karl Dubost when he asked about that:
Chris and I are still deciding what to do with it. The draft is
good (technical), but the reaction (political) was generally
apathetic which was actually a touch better than expected, but
still not best-case which is what we were aiming for. The only
things that really need changing are some excellent schema
improvements that were suggested by Eric VDV [4].
We could have had several thousand feeds, for example. That wouldn't
be much even compared to the penetration that RSS 1.0 has had, and it
wouldn't have done any harm. The benefit would've been that anybody
who disliked the original specification as much as we did could've
slept better at nights. That's all.
Dan Brickley said last year [5] that he's "waiting to see how Atom
goes, before worrying too much about RSS1.x futures". He was wondering
more about the protocol, but in that time I've come to decide that I
quite like Atom Syntax. It's relatively cruftless, architecturally
solid, and people agree on it which is the main thing. And it works,
of course, but that goes without saying. I don't particularly care
that it's not RDF based. I still rather like the largely ignored
Extensibility Framework [6] that Ken MacLeod and Sjoerd Visscher
rustled up with me, but that's fine. I still doff my cap to Sam Ruby.
Bjoern Hoehrmann recently, comically, suggested [7] that what we
really need is an RSS 1.0 v2.0. If he'd've said *needed*, he would
have been mysteriously right: if Chris and Cody and I had worked on
purely bugfixing RSS 1.0, and doing so slowly and collaboratively, we
might've gained a lot more traction. But that would have been no fun.
I am sad that I didn't get a chance to implement Eric VDV's schema
suggestions, but on the other hand there is simply no point unless
anyone wants to make heavy use of the 1.1 specification. And I don't
believe that such interest will ever be generated now, notwithstanding
the distributed cabal of people that are tentatively interested in
such a thing. Just use Atom instead.
So that's it. RSS 1.1 stays as it is, and I'm happy with it, and I'm
happy with what we did. I don't recommend that anyone else try to do
the same thing, though, either by working on RSS 1.0 or 1.1 v2 or RSS
1.2. Why? Because for my misdemeanor of creating a new syndication
language, I later killed one--Avida's RSS 3.0 [8]. You might not have
the same opportunity.
[1] http://groups.yahoo.com/group/rss-dev/message/6903
[2] http://inamidst.com/rss1.1/
[3] http://chatlogs.planetrdf.com/swig/2006-03-28.html#T07-20-27
[4] http://groups.yahoo.com/group/rss-dev/message/7018
[5] http://chatlogs.planetrdf.com/swig/2005-04-04.html#T20-39-17
[6] http://inamidst.com/atomef/
[7] http://chatlogs.planetrdf.com/swig/2006-03-28.html#T07-32-58
[8] http://lists.w3.org/Archives/Public/www-archive/2005Aug/0018
--
Sean B. Palmer, UK, 2006
http://inamidst.com/sbp/
Yahoo! Groups Links
<*> To visit your group on the web, go to:
http://groups.yahoo.com/group/rss-dev/
<*> To unsubscribe from this group, send an email to:
rss-dev-unsubscribe-***@public.gmane.org
<*> Your use of Yahoo! Groups is subject to:
http://docs.yahoo.com/info/terms/
On 18th January 2005, Christopher Schmidt and I, with the kind support
of Cody Woodard, released [1] a draft of RSS 1.1 [2] to this mailing
list. Every few weeks or so, either Chris or a member of the public
enquires, "what's the future for RSS 1.1? Why has it stopped being
developed?" (e.g. [3]), which I'm going to answer here.
RSS 1.1 was born of the fact of us liking RSS 1.0 but recognising that
it had many bugs, and a poor specification. I know many of the authors
of 1.0, and some of them are close personal friends; indeed, I learned
much of what I know about writing specifications from Aaron Swartz. I
have tremendous respect for them as writers of specifications, so I'm
not fully sure how RSS 1.0 came off so badly. Some of it was technical
limitation: I had recourse, for example, to RELAX NG Compact, which
the RSS WG did not. As for the rest, I'm too scared to look through
the archives to find out.
So when Chris and I expressed our mutual frustrations with 1.0 at one
another, we decided that rather than lauch yet another stonking
diatribe against the quality of the RSS 1.0 spec, as Bijan Parsia once
memorably and anyway unbeatably did, we thought we'd avoid the route
of the unmarried marriage counsellor and simply write a new
specification ourselves. And as a specification, 1.1 turned out to be
quite passable. Look kids, it has a conformance section!
We didn't know what to expect when we released the draft, but we had
an impression that we'd get the usual range of good, middling, and
ugly responses. In fact, we got far more ugly responses than I'd
expected--some quite considerably incompetent stuff to the point where
I gave up and let Chris deal with filtering it all.
On the good side, we got more implementations than expected, and a few
people embracing the new version. But the best of the feedback came
from the middling responses--the people who commented merely on the
technical issues of the specification and didn't bother to comment on
the politics of it. All of the bad responses that we got were either
political, or based on misfounded technical assumptions and ideas. It
all made me rather glad that I speak to people like Bjoern Hoehrmann
and John Cowan every day.
Chris and I wrote the specification and the validator and test suite
and all the implementations in about three days. The entry barrier for
producing a syndication language is not high, because getting it to
just work is not rocket science. That's why there are sundry versions
of RSS. Getting it to work perfectly, on the other hand, is a
Will-o'-the-wisp, and one which the Atom folk may be forever chasing,
though they're certainly the closest to it.
Anyway, the future. As I said to Karl Dubost when he asked about that:
Chris and I are still deciding what to do with it. The draft is
good (technical), but the reaction (political) was generally
apathetic which was actually a touch better than expected, but
still not best-case which is what we were aiming for. The only
things that really need changing are some excellent schema
improvements that were suggested by Eric VDV [4].
We could have had several thousand feeds, for example. That wouldn't
be much even compared to the penetration that RSS 1.0 has had, and it
wouldn't have done any harm. The benefit would've been that anybody
who disliked the original specification as much as we did could've
slept better at nights. That's all.
Dan Brickley said last year [5] that he's "waiting to see how Atom
goes, before worrying too much about RSS1.x futures". He was wondering
more about the protocol, but in that time I've come to decide that I
quite like Atom Syntax. It's relatively cruftless, architecturally
solid, and people agree on it which is the main thing. And it works,
of course, but that goes without saying. I don't particularly care
that it's not RDF based. I still rather like the largely ignored
Extensibility Framework [6] that Ken MacLeod and Sjoerd Visscher
rustled up with me, but that's fine. I still doff my cap to Sam Ruby.
Bjoern Hoehrmann recently, comically, suggested [7] that what we
really need is an RSS 1.0 v2.0. If he'd've said *needed*, he would
have been mysteriously right: if Chris and Cody and I had worked on
purely bugfixing RSS 1.0, and doing so slowly and collaboratively, we
might've gained a lot more traction. But that would have been no fun.
I am sad that I didn't get a chance to implement Eric VDV's schema
suggestions, but on the other hand there is simply no point unless
anyone wants to make heavy use of the 1.1 specification. And I don't
believe that such interest will ever be generated now, notwithstanding
the distributed cabal of people that are tentatively interested in
such a thing. Just use Atom instead.
So that's it. RSS 1.1 stays as it is, and I'm happy with it, and I'm
happy with what we did. I don't recommend that anyone else try to do
the same thing, though, either by working on RSS 1.0 or 1.1 v2 or RSS
1.2. Why? Because for my misdemeanor of creating a new syndication
language, I later killed one--Avida's RSS 3.0 [8]. You might not have
the same opportunity.
[1] http://groups.yahoo.com/group/rss-dev/message/6903
[2] http://inamidst.com/rss1.1/
[3] http://chatlogs.planetrdf.com/swig/2006-03-28.html#T07-20-27
[4] http://groups.yahoo.com/group/rss-dev/message/7018
[5] http://chatlogs.planetrdf.com/swig/2005-04-04.html#T20-39-17
[6] http://inamidst.com/atomef/
[7] http://chatlogs.planetrdf.com/swig/2006-03-28.html#T07-32-58
[8] http://lists.w3.org/Archives/Public/www-archive/2005Aug/0018
--
Sean B. Palmer, UK, 2006
http://inamidst.com/sbp/
Yahoo! Groups Links
<*> To visit your group on the web, go to:
http://groups.yahoo.com/group/rss-dev/
<*> To unsubscribe from this group, send an email to:
rss-dev-unsubscribe-***@public.gmane.org
<*> Your use of Yahoo! Groups is subject to:
http://docs.yahoo.com/info/terms/