No hygienic macros. about husk-scheme HOT 13 CLOSED

Eli - Thanks for your time, I'll look into both of these issues.
I need this kind of input to improve the project...

from husk-scheme.

15 minutes ago, justinethier wrote:

Eli - Thanks for your time, I'll look into both of these issues.
I need this kind of input to improve the project...

Well, FWIW, I suspect that such things would require some major
revisions...

      ((lambda (x) (x x)) (lambda (x) (x x)))          Eli Barzilay:
                http://barzilay.org/                   Maze is Life!

from husk-scheme.

That's fine, if the changes aren't feasible then at least there's a record here. Thanks again for your feedback!

from husk-scheme.

Version 3.1 (now in the master branch) fixes issues with lexical scope that were noted in this ticket.

from husk-scheme.

http://srfi.schemers.org/srfi-72/srfi-72.html

from husk-scheme.

Yeah, that's an explanation of terms around the area of a syntax-case macro system. Implementing that would be good, but challenging.

from husk-scheme.

"Common Lisp went with the pragmatic decision that the macro writer is responsible for the 'hygiene'", this is correct only for the easy side of hygiene that is solved with a gensym. The other side (where bindings in the use site shadow bindings in the macro definition site) is hard to solve and requries a global solution.

See: http://tmp.barzilay.org/tutorial.txt | Breaking Hygiene

Specifically:

Hygienic macros avoid variable capture of two kinds:

• names that are introduced by the macro do not clash with names in
  sub-expressions that are passed from user code to it -- for example,
  we use temporary variable names without worrying whether
  subexpressions use the name,
• names that the macro references are coming from the lexical context of
  its definition, rather than the lexical context of its use. For
  example, if a macro uses if', then the definition of if' that was in
  effect at the context where the macro is defined is used.

Clinger and Rees present an algorithm to solve both capture problems in Macros That Work:
http://mumble.net/~jar/pubs/macros_that_work.ps

from husk-scheme.

That work is pretty old -- there was a lot since then. IMO, it's best to read about the syntax-case macro system, if you want to do the right thing.

from husk-scheme.

You make a good point, syntax-rules is older and does not support low level macros. But it is also part of
R⁵RS and R⁷RS (small language draft), which husk is targeting, while syntax-case is not. I need to walk before I can run on this. I'll add another ticket to include syntax-case in a future release...

from husk-scheme.

Hygienic macro support has been added to the husk 3.4.x series using syntax-rules. Both "sides" of hygiene are included, and husk now passes all tests from this ticket as well as many more from the test suite. Macros will continue to be tested and refined before this ticket is closed out.

from husk-scheme.

This work is essentially complete except an issue with define noted here: #65

I am interested if you have any other test cases for macros that are not transformed correctly... otherwise this issue is going to be closed out soon.

from husk-scheme.

I'd play with it more, but dealing with Haskell is difficult for me ATM.

from husk-scheme.

No problem - I am closing this ticket since hygiene is implemented now, minus an outstanding issue or two that are already being dealt with under separate tickets. If you ever do have a chance to double-check and see something that is not working, please let me know.

Thanks for your time.

from husk-scheme.