Jens Axel Søgaard:
> For the record infix.plt was not meant to be the final say in infix notation.

Fair enough.  But infix.plt is a reasonable effort, it's widely available, and it's a fair representative of other infix approaches too.

> The implementation on Planet was just meant to be something to try out.
> In fact I have changed some of the decisions in Bracket.
>
> Since this discussion is in the rationale section, I think, the rationale
> of infix.plt is more relevant than the actual implementation.
...

Okay, but for detailed comparisons we really do need the details. So I propose that we *add* your (very nice) higher-level rationale for infix.plt, before the details we currently have.  That should make it easier to follow the syntactic details about infix.plt that then follow.

> I think the comment about the grammar in the source should
> be left out. It wasn't meant to be used as final documentation.
> It was just to get started. Read the actual source to see
> the exact accepted grammar.

I *did* read the source code, but most SRFI-105 readers won't want to wade through it.  For discussion purposes, what we need is the infix.plt grammar, but there is no one place where the complete infix.plt grammar is concisely defined.  So I derived the infix.plt grammar by starting where MOST of the grammar is defined, namely the front of parser.ss, and I give a full citation/credit for that.  I then tried to summarize those omissions (by examining the documentation and source code) to complete the grammar.

The "comment about the grammar in the source" is a *citation* of the source of information.  We shouldn't remove that text, because it's important that we give credit.

Thanks again for your comments!  I'll soon post my latest attempt at responding to your comments.  I can't promise to make everyone happy, but if there's an *error* I certainly want to know.

--- David A. Wheeler