On Wed 15 Feb 2012 18:50, Mark H Weaver <m...@netris.org> writes: > Here's another patch set to improve support for source properties.
LGTM in general. A few questions to address now or later; I'm fine with pushing as is. On the broad side, does the current "source-properties" mechanism make sense? I get the feeling that a variant of `read' that returned annotated objects would be better in many ways. Perhaps that is an orthogonal question though. For prior art, see Racket's read-syntax: http://docs.racket-lang.org/reference/Reading.html#(def._((quote._~23~25kernel)._read-syntax)) > * add the 'supports-source-properties?' predicate (as well as > scm_supports_source_properties_p), which cannot be implemented > efficiently in Scheme. What do you think about a name like `unique-datum?'? It's uniqueness (in the eq? sense) that is the thing, right? > * psyntax now accesses and sets source properties for all supported > objects. Previously it assumed that only pairs could support source > properties. Do check to see that the result of ,time (compile-file "module/ice-9/psyntax.scm") is not too badly affected. Cheers, Andy -- http://wingolog.org/
