quale@saavik.cs.wisc.edu (Douglas E. Quale) (04/06/91)
When I posted a suggestion that attempts to evaluate () should be required to signal an error, I didn't expect a very interesting discussion of multiple values to arise (via via return values for set! et al via order of evaluation). The () vs #f issue is old and tired by now. Looking at the IEEE Draft Standard P1178/D3 (from Nov. 18, 1989, now a year and a half old), I would like to know what the standards committee and the Scheme community in general see as the future of I/O in Scheme. The Draft eliminates char-ready? found in R3RS. How can a read be performed without fear of hanging? As an odd note, Scheme implementations should be encouraged not to line buffer interactive output, or else should flush the output on reads. Line buffering interactive output requires prompts to end with newline, which isn't very pretty. As a last thought, the Draft requires type disjointness that no longer allows chars to be numbers. Now write-char can safely be replaced by display. (The name write-char is somewhat of a misnomer. The effect is actually what one would expect of display-char.) -- Doug Quale quale@saavik.cs.wisc.edu