Path: utzoo!utgpu!watserv1!watmath!att!pacbell!pacbell.com!ames!ucsd!usc!zaphod.mps.ohio-state.edu!swrinde!cs.utexas.edu!uunet!willett!dwp From: dwp@willett.UUCP (Doug Philips) Newsgroups: comp.lang.forth Subject: Re: ANS FORTH TECHNICAL COMMITTEE Message-ID: <1075.UUL1.3#5129@willett.UUCP> Date: 3 Jun 90 22:46:18 GMT References: <1069.UUL1.3#5129@willett.UUCP> Organization: Latest link in the ForthNet chain. (Pgh, PA) Lines: 68 In <1069.UUL1.3#5129@willett.UUCP>, D.RUFFER [Dennis] writes: > Re: ir230@sdcc6.ucsd.edu (john wavrik) > > > Declaring the action of fundamental words to be "implementation > > dependent" is not the kind of compromise which will contribute to > > portability. > > John, that wording is only used in cases where the committee members can not > agree on the action. IMHO, specifying where those grey areas are goes a long > way to helping me decide what things to avoid when writting a "standard > application". It also means that I can choose to take advantage of a vendor's > "value added" features if I do not care about portability. I agree with Dennis here. One of the most useful things that came out of X3J11 was a classification of kinds of behaviours into which things would be put. Implementation dependant is certainly a valid one, so long as the entire standard doesn't end up in it! You need to know exactly what is *and isn't* guaranteed by the standard. Knowing that word is implementation defined means that you'll avoid using word (or will explicitly define it yourself) if you care about portability. Just because this bin doesn't say strict things about the semantics of a word doesn't mean the bin is useless. (see Plauger's column in Computer Language [early 1990's, I can get the specific issue if anyone is interested] for more info along these lines). > not sufficient. There are many people who can not contribute. However, > anyone can order a copy of the BASIS, and starting with BASIS 12, anyone will > be able to download it in electronic format. Side Issue: Is this something that will benefit other standards or not? How did X3J14 pull it off? (X3J11 never did, but one doesn't know how hard they tried either, or what the political changes since then have been?) > Suppose you were required to write an application that > would run on any ANS Forth. Would it be possible? > > I also removed the requirement that it be as easy > to write the application as it is now. Certainly writting portable > applications is going to be harder than it is to write non-portable ones. > Anyone who believes otherwise has not been writting applications long enough. Yep. There was lots of flap about this kind of thing with X3J11. "If you do that you'll break my application/user's application." Such people don't seem to realize that ANSI/X3/X3Jn doesn't have that kind of power. Pointing out that something isn't portable doesn't suddenly make go from being portable to being not portable! On the other hand, there will be no X3J14 police going around to make sure your Forth conforms. If you want to push the envelope, the ANSI standard won't stop you. There is a case for having "standard libraries" even if they aren't part of "Forth" itself, or of the standard. I'm not quite sure how that worked out for the C people, but it makes a lot of sense. I'm not talking about layers of standard here, but just "appendices", such as Appendix J, the wordset. I think this is a good kind of thing and hope it manages to stay in. (It ain't over till its over!) > Let's just hope > that the process does not kill the evolution. It probably won't kill the evolution, but it might just take out a few people! :-) -Doug --- Preferred: willett!dwp@hobbes.cert.sei.cmu.edu OR ...!sei!willett!dwp Daily: ...!{uunet,nfsun}!willett!dwp [in a pinch: dwp@vega.fac.cs.cmu.edu]