Shebs%UTAH-20@sri-unix.UUCP (07/10/84)
From: Stan Shebs <Shebs@UTAH-20> YES! Although I don't care much for the idea of a transport syntax parser written in C or Pascal (some of us are trying to avoid those low-level languages altogether!). Why not set up a DCG for the transport syntax, then have implementors translate to the syntax for their implementation? A version in DEC-10 Prolog could be supplied as an example, to be hacked as necessary. The problem I see with a C version is that it will be horribly gross and almost certainly incorrect. Since Prolog is such a wonderful language, it should be quite easy to take the transport syntax DCG, produce a parser, and load it on top ofan implementation, and either use it directly in a "compatibility mode", or compile code in transport syntax into the implementation's syntax. Aside: most Lisps can be set up to read each other's syntax with only a small amount of work on readmacros, etc. In fact, PSL's reader is built using a YACC lookalike and rules for parsing s-expressions, so *none* of the syntax is wired in. Another aside: I don't like CGOL. Actually, I don't care much for Interlisp as a whole; it's a prime example of grotesque hackery that Prologers would do well to beware of (experience has already shown that nasty Prolog hacks are just as prevalent as in any other language)... -- Stan Shebs
O'KeefeHPS@sri-unix.UUCP (09/09/84)
From: O'Keefe HPS (on ERCC DEC-10) I have now patched all the holes I could find in my previous proposal on this topic, written it up, and given some examples. By the time you read this, the paper should be available in the Prolog library (as TRANSPORT.MSS). If you care about moving code between different Prolog dialects, please read this paper and send your comments to me or the Digest or both. The paper is just my opinion, and if you think there's a gaping hole in it you may well be right. To be of any use, we have to agree about transport syntax, and now is the time to do it. Don't take the fact that the transport syntax looks vaguely like LM-Prolog as an endorsement of anything but DEC-10 syntax for human consumption. The transport syntax I've come up with is (I think) trivial to parse, makes all the necessary distinctions, and has a hundred other virtues, but I don't **like** it.