As a long-time standards user  and contributer (in the data
communications field) my
opinion is that there is no completely 'right way' to do a standards
process.  If there is a need for a function, it will get done lots of
different ways before a standard is finally agreed.   We can learn
from history, by looking at two 'somewhat competing' standards
bodies and what was achieved. (Note that the details may not be
right, but you get my meaning)

The International Standards Organization (ISO) created a bunch of
standards
for data communications which were a marvel of architectural rigour and
elegance - anybody remember the Open Standards Interconnect 7-layer
model.  Unfortunately, they were cumbersome to implement and nobody
wanted to be first because they were in a state of flux.  In particular
the
network management standards were very difficult to understand.

Meanwhile a bunch of yahoos took the most import parts of the OSI
network
management standards and implemented a simple network management
protocol (SNMP) that could be quickly implemented by anybody.  Once
it was working they submitted it as a draft Request for Comments (RFC)
to
the Internet Engineering Task Force, and it quickly became the standard
most network devices use today.  FOR ALL ITS FAULTS, and it has quite
a few, it was good enough to do the job and implementable  quickly.
Today,
nobody cares about the OSI network management protocols.

New versions  of an IETF standard must be accompanied by a reference
implementation, so you don't get the case where a standard is not
implementable.

To draw an analogy, we need to
- issue the equivalent of an IETF RFC that describes the current
standard and
  nominate a reference implementation.  This means we need to freeze one
of
  the open-source implementations at that level, and ensure that other
implementations
  support it.  A test suite would be a really nice thing to have, too.

- issue a reference implementaton and specification for the next version
of abc (which,
  by the way, does not need to support everything we've talked about so
far, only
  the parts we agree on)

Meanwhile, extensions are perfectly fine in anybody's software, but they
should
be identified as such.  Eventually they become candidates for the
standards process.

By the way, I still haven't implemented the V: directive in Skink
because I can't
decide which of the several competing approaches is most likely to
survive. Phil's
recent work in reformulating the same tune to make it compatible over
more packages
gives me hope that we can come up with a common format.



[EMAIL PROTECTED] wrote:

> Phil Taylor said -
>
> >1.  Anything but the most simple extension needs some experimentation
>
> >to find out what works.  You've got to do it first, then try it out
> with
> >a lot of music to see if it's a good idea.
>
>
> I don't see why that precludes discussing it on the list before
> releasing it.
>  At the very least, that would mean people were aware of what was
> being done.
>
> >2.  If we had to wait for agreement nothing would ever get done.
>
> So it gets done whether there is agreement or not.  Does this mean you
>
> wouldn't object to me introducing a sharps/flats and no tonic version
> of the
> K: command?
>
> >Nobody's hostile to anyone here (present company excepted, of
> course).
>
> Well, Jim Vint seems to come in for some fairly robust criticism and
> as for
> me, at least you are responding to what I say rather than going
> straight into
> "Bryan bashing mode" .  That's encouraging.
>
> >middle = was proposed after the draft standard appeared.
>
> So put up a new proposal.  You are on the standards committee, dammit.
>
> >I never bothered proposing Gregorian chant as a general standard as
> it's
> >such a specialised area.
>
> Fair enough.  Nobody is going to be accused of an incomplete
> implementation
> of abc if they don't include it in their software.  On the other hand,
> if it
> was a published standard, people could take it up if they wanted and
> ensure
> that their version was compatible with yours.
>
> >>>Multivoice abc using V: started in abc2midi and (at least in its
> basics)
> >>>is now supported by most programs.
> >>
> >>In at least two incompatible versions (see above).
> >
> >Not completely incompatible (see above).
>
> That sounds a bit like "slightly pregnant" or "almost a virgin".  Jack
> Campin
> posted an interesting version of Shingly Beach recently using BarFly
> notation.  Muse, abcmus and my own abc2nwc couldn't cope with it.
> abc2win
> produced all the right notes but not necessarily in the right order.
> I don't
> imagine abc2ps could handle it.  Frank Nordberg recently posted
> separate
> BarFly and abc2ps versions of a tune.
>
> >The draft standard is just that.  It represents one person's
> proposals
> >put up for discussion.  I've made my opinion on that section clear on
>
> >several occasions, and will continue to do so.
>
> It actually represents something that someone has already implemented
> and
> then put up for discussion; a procedure that you endorse in your point
> 1
> above.  Once somebody has put a lot of work into implementing
> something they
> think is a good idea, they aren't going to throw it all away because
> somebody
> then says they don't like it, as I'm sure you'll agree.
>
> One day there will be a single agreed abc standard (maybe, but not
> yet).
>
> Bryan Creer
>
To subscribe/unsubscribe, point your browser to: http://www.tullochgorm.com/lists.html

Reply via email to