A few comments/corrections on what I said before about R7RS... > The draft R7RS says for equal? "Even if its arguments are circular data > structures, equal? must always terminate." That sounds nice and general, > except what that MEANS is that every test using "equal?" must be ten times > slower and chew up memory.
You to partly solve this for equal? by using a counter to recursively implement "equal?" over pairs up to some value, and if you exceed the value, switch to a slower equal test that tracks every comparison. You still have twice as many comparisons, and you still have to track in the worst case. Also, R7RS doesn't have hash tables built in, so there's no easy way to self-implement an efficient implementation in a portable way. It still seems to me that an equal-circ? or similar would be better. This is a big change to the semantics, in particular, it *requires* a much more complex and memory-hungry algorithm. It seems to me it should have a different symbol. It still creates a portability problem; if they'd named the new operator "equal-circ?", then implementations or app writers could ADD that without interfering with the existing equal? > Same for write. The write is more complicated. R7RS has a "write-simp", and have *CHANGED* the semantics of write so that the written output can't be read by older systems. Eeek; we intentionally talk about NOT changing the writer for even curly-infix for years to come, just to avoid that issue. I think this should be write-circ, and keep using write as is. The good news is that at least some of the problems with "write" are known: http://trac.sacrideo.us/wg/wiki/WG1Ballot6Results Looks like they're going to have write detect cycles, and separately provide 'write-simple' and 'write-shared'. That still means a significant overhead in "normal write" to track every cons pair seen :-(. > (as well as an appeal to restored the c....r functions which are some of the > few actually portable functions and have been with us for decades). Correction: Removing them was discussed here: http://lists.scheme-reports.org/pipermail/scheme-reports/2011-October/001575.html http://lists.scheme-reports.org/pipermail/scheme-reports/2011-October/001587.html But cadddr and friends *are* in the latest draft 6. Good. It's nice to have SOME procedures that are common between versions of the language :-). I've read a lot of R7RS discussions recently, and I got confused between what was discussed and what was agreed on. Sorry about that! --- 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
