Re: some questions about pragmas, and the arguments as I understand them

Innovimax W3C writes:

> Hello all,

> Hope all of you are well !

And similarly I hope that all is well with you!  It is a pleasurable
surprise to hear from you on this list.

> Thank you very much Michael, it's helping a lot to get the depth of
> situation.

> Just to add perhaps a little more salt to this discussion, il would like to
> propose to eat our own dog food

> What about allowing only one type of comment and let people define some
> ixml grammar on how to process those comments to extract "pragma",
> "localisation", "author" or whatever idea they can come up with

I think that that would amount to a qualified "no" on question 4:  No,
ixml should not provide distinct constructs, with the explanation that
individuals or groups can use ixml to define grammars for specific types
of information.

I would formulate the arguments for and against this roughly as follows:

  for:

    - It demonstrates the power of ixml to organize information, and the
      usefulness of putting the power to write grammars in the hands of
      users instead of reserving that power to standards groups.

    - (non-technical argument) It removes a contentious issue from the
      CG and lets it be worked out outside the group.
      
  against:

    - It has the same interoperability issues as any other 'No' answer
      to question 4:  since multiple grammars for various kinds of
      inline out-of-band information may be defined, it will be possible
      for a grammar prepared using one such grammar to be submitted to a
      processor which uses a different such grammar.

      It thus increases the likelihood that the a grammar that works
      successfully with one processor will produce the same results with
      others.

In practice, however, I think it is quite likely, if the ixml spec does
not define a construct for pragmas, that individual implementations
which need some form of inline out-of-band signalling will define
grammars for those pragmas.  It is also likely (though I think perhaps
less certain) that grammars for pragmas in general may be published and
implemented -- not grammars for specific pragmas, but for a class of
pragmas distinct from comments not directed to a processor.

The only question is whether those who specify such grammars will limit
themselves to using the existing 'comment' construct of ixml, or will
take seriously the argument that pragmas should not be used to do things
that belong in the base language, and use that argument to justify
incompatible extensions or modifications of the ixml specification
grammar.

There are people who believe that language designers do not face a
choice between a language which can be extended and a language which
cannot be extended, but only a choice between a language with defined
extension mechanisms, which can be identified as such, and a language in
which any extension will involve an incompatible change to the syntax.
In the software ecology this is a choice between 'controlled' extension
(for some definition of 'controlled' -- perhaps a better term would just
be 'clearly identified' extension)) and uncontrolled extension.
      

> Hence we provide only comments without impact on the actual processing of
> the grammar, and people can standardise at their speed the way they want to
> use the "semantics" of those comments by working on another ixml grammar

> My two cents

For which I thank you.

-- 
C. M. Sperberg-McQueen
Black Mesa Technologies LLC
http://blackmesatech.com

Received on Thursday, 3 February 2022 14:53:18 UTC