On Tue, Mar 29, 2011 at 4:43 PM, B Smith-Mannschott wrote: > Horrible hack, maybe, but it got me thinking. What you seem to be > doing is moving between "code" and "literal" mode by quoting with #. > This is a little like traditional quasi-quote... > > And that got me thinking about Scribble [1] again. In this > context I think of scribble as being sort of an inverse of normal > scheme syntax. In the end, the scribble reader produces the same > kind of data structures as the normal scheme reader, but the > emphasis is moved from code to textual content. Source is content > by default, but you can escape into logic.
This is kind of a good summary of the reader's intended use, beyond our documentation system: it's fitting for any kind of code that deals with a lot of textual content -- and having a kind of cheap DSLs for interacting with other tools by generating source is exactly one of the things that you can do with it. But getting that from the scribble documentation will be a little hard -- I wrote a paper on just the implementation and use cases of the reader that will be much more fitting in this context, with some examples of such uses: http://barzilay.org/misc/scribble-reader.pdf On Mar 29, 7:56 pm, Ken Wesson <kwess...@gmail.com> wrote: > TeX and LaTeX work that way, too. And in a certain sense so does > syntax quote in Clojure itself, where the macro expansion source is > the "content" and unquoted expressions can fill bits in by "escaping > into logic". Yes for TeX and LaTeX, but that's obfuscated by the programming model that makes it necessary (since macros "call" other macros by producing text with those macros). But IIUC, Clojure uses a traditional lisp-like quote mechanism, and Scribble is different from that in an important way. The problem with quasiquotes is that you need to be aware of the context where code is written -- you need to explicitly unquote code that is intended to run, and that's hard and error prone. (My favorite example is an old version of the PLT Scheme pages that had lots of "<quote>nbsp</quote>"s because 'nbsp was used in a context that was already quoted.) In contrast, the Scribble reader's approach is that all scribble forms are, in a sense, "unquoted". This wouldn't be as convenient for the kind of lispish code generation that you do with quasiquotes -- roughly speaking, you'd need to quote each head form that is intended to produce code -- but it's extremely convenient for dealing with text. (This is actually based on a way of reflecting logical terms in a theorem prover I worked on...) In any case, that reader paper describes all the decisions that lead to this design, and how it works in providing a better facility than here-strings and string-interpolation, with lots of examples. (And in case it wasn't clear -- I strongly believe that this kind of syntax is useful in any language, but much more in lispish languages since they already enjoy the huge benefits of a uniform representation of sexprs.) -- You received this message because you are subscribed to the Google Groups "Clojure" group. To post to this group, send email to clojure@googlegroups.com Note that posts from new members are moderated - please be patient with your first post. To unsubscribe from this group, send email to clojure+unsubscr...@googlegroups.com For more options, visit this group at http://groups.google.com/group/clojure?hl=en
