Handling #|...|# creates a lot of implementation complexities and otherwise-unnecessary interdependencies. Handling #|...|# has multiple changes all the way through the code that are otherwise unnecessary, and create easy opportunities for bugs. The same is true for #!...!# if they have similar semantics.
Only the sweet-expression reader really needs to handle #|...|# specially. We could create a special routine that calls the neoteric-reader, but handles #|...|# specially instead, isolating the code more. Implementation methods are all ugly that I see; they include: 1. Peeking two characters ahead. Not portable in either Scheme or Common Lisp at *least*. 2. Modifying the underlying hash table for #|...|#, and then having the neoteric-reader modify the table again on recursion. Basically, the readers constantly manipulate the table. (Better not have parallelism!) Yuk. A *much* simpler approach would be to change the spec so it's trivial to implement. PROPOSAL: Change the semantics of #|...|# for the sweet-reader to: "All whitespace and ;-comments after multi-character comments are consumed; the datum is considered to begin where the comment began. This is because a #|...|# comment is read recursively by the neoteric reader." Under this semantic, this would work unchanged: #| hello |# define x 5 And this would also be unchanged: define x #| Clever thought goes here. |# 5 This *WOULD* be changed; it becomes "define x 5": define x #| Whitespace and comments after multi-line comments are ignored |# ; This is ignored too, because the neoteric-reader is trying to find something to read. 5 Perhaps this is less "pure", but it seems to me that people shouldn't normally do that anyway. As long as we clearly define the result, I think we're okay. The result would be a MUCH MUCH simpler and cleaner implementation. The reader implementations would go back to being simple and not interdependent. I think simple implementation is important; we want to get this widely adopted, and anything that makes it hard to implement may slow adoption. And I don't think supporting the current semantic is really that important, certainly not enough to create the complicated implementation. Thoughts? Comments? --- David A. Wheeler ------------------------------------------------------------------------------ Live Security Virtual Conference Exclusive live event will cover all the ways today's security and threat landscape has changed and how IT managers can respond. Discussions will include endpoint security, mobile security and the latest in malware threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/ _______________________________________________ Readable-discuss mailing list Readable-discuss@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/readable-discuss
