Notes on Syntax

[David Nolen]

http://olabini.com/blog/2012/02/notes-on-syntax/

Interesting read. Somewhat related, what do people think about the VPRI 2011 Report http://www.vpri.org/pdf/tr2011004_steps11.pdf? Will the construction of many different languages become an essential skill? Racket's Languages as Libraries http://www.ccs.neu.edu/scheme/pubs/pldi11-thacff.pdf seems a different take on this idea?

[Peteris Erins]

To support the third paper and borrowing from Mathematics, the ability
to establish local syntax is key. Mathematics has many layers of
subsyntax, e.g., fields, research groups, papers, down all the way to
even individual lemmas. Adoption of Mathematical notation seems to be
more faithful. In comparison, an individual macro will almost
certainly be tweaked with and "improved" unless it becomes part of an
implementation or a whole DSL that people are less likely to touch.

--
Peteris Erins

On Feb 12, 1:16 pm, David Nolen <dnolen.li...@gmail.com> wrote:
> http://olabini.com/blog/2012/02/notes-on-syntax/
>
> Interesting read. Somewhat related, what do people think about the VPRI

> 2011 Reporthttp://www.vpri.org/pdf/tr2011004_steps11.pdf?Will the

> construction of many different languages become an essential skill?

> Racket's Languages as Librarieshttp://www.ccs.neu.edu/scheme/pubs/pldi11-thacff.pdfseems a different take
> on this idea?

[Þórarinn Sigurðsson]

A few thoughts: Would the "domain-optimized" approach in the VPRI
report make it more difficult for the various POLs to talk to each
other when compared, for example, with the "languages-as-libraries"
approach? In the latter case, the obvious solution might be to
standardize on just using the most established data structures/types
of the "father language" (e.g. scheme) to communicate; in that case
there is a good deal of common ground that could make composing
constructs written in sibling sublanguages easy.

How would this work with the POL approach? A compiler/interpreter that
understands all the POLs and is able to handle interlingual
communication behind the scenes? Would this hamper the addition of new
POLs due to the explicit synchronization burden on the compiler
writers? When writing a DSL in a Lisp, the author has the burden of
understanding Lisp. Does this mean that in the VPRI approach, people
who want to write new POLs have to understand the entirety of the
underlying "common back end"?

I'm wondering if it's a tradeoff between better-optimized syntax
(VPRI-approach) versus easier extensibility and composability
(language-as-a-platform).