Re: A few random I/O proposals

[david carlton <carlton@husc.harvard.edu>]

On Tue, 26 May 92 17:49:59 -0400, "Guillermo J. Rozas" <jinx@martigny.ai.mit.edu> said:

[ Talking about sending output to strings. ]

> There are other possibilities.
> One of them would be to include the pair of procedures
> make-string-output-port and string-output-port-content.

> MIT Scheme has

>   (with-output-to-string thunk)

> and

>   (with-output-to-truncated-string max thunk)

> with-output-to-string returns a string.
> with-output-to-truncated-string returns the cons of a boolean and a
> string.  The boolean indicates whether the output was in fact
> truncated (true if it was truncated, false if it finished).

> It has string->input-port and with-input-from-port for input.

I think the former is probably a better idea, at least if there is
only going to be one way in the standard for sending output to
strings.  I can imagine a situation occuring where somebody might want
to send output to two separate places at once, and
with-output-to-string doesn't cover that.  At any rate, they're both
better than the possibilities I mentioned.

Minor points: I would call make-string-output-port open-output-string
instead, and string-output-port-content output-string-contents.
Output-string-contents should work on a port that has been closed.  I
wouldn't want to include the with-output-to-truncated-string, and
would call string->input-port open-input-string instead.  Also is
with-input-from-port a typo?  Whether or not it is, having a generic
with-input-from-port (and with-output-to-port) might not be a bad idea
- much better than having to add an extra pair of procedures for each
new kind of port that a given implementation supports.

david carlton
carlton@husc.harvard.edu

       Concentrate on th'cute, li'l CARTOON GUYS!  Remember the SERIAL
       NUMBERS!!  Follow the WHIPPLE AVE EXIT!!  Have a FREE PEPSI!!
       Turn LEFT at th'HOLIDAY INN!!  JOIN the CREDIT WORLD!!  MAKE me
       an OFFER!!!