On 2009-09-08, at 13:02, Alaric Snell-Pym wrote:
I think that most of the effort in porting something between
implementations is unnecessary.

Now, if two Schemes had taken opposite stances in some deep
philosophical debate about, say,  how hashtables should be modelled,
then I'd be... not exactly ecstatic, but unconcerned about them having
different models. I'd see it as a temporary affair, until the great
"Is A Hashtable More Like An Alist Or A Procedure" (or whatever) war
subsided, and we ended up with one or the other winning, or a sensible
way of merging the two views into one appeared.

But when they have different ways of accessing the network, different
ways of opening a subprocess to run a command, and different ways of
structuring my code into separate files and expressing the
relationship (so even if I wasn't using such platform-specific
features, I'd have unportable code), then it feels like unnecessary
work doing the porting.

Agreed, but I long ago learned to create tiny middleware libraries to ease porting problems. So we could paper over the incompatibilities that way, and thus remove ease of porting as a goal of R7RS. Having said that, gratuitous incompatibility helps nobody.

-- v
_______________________________________________
r6rs-discuss mailing list
[email protected]
http://lists.r6rs.org/cgi-bin/mailman/listinfo/r6rs-discuss

Reply via email to