Path: csiph.com!v102.xanadu-bbs.net!xanadu-bbs.net!news.glorb.com!news-out.readnews.com!news-xxxfer.readnews.com!postnews.google.com!glegroupsg2000goo.googlegroups.com!not-for-mail From: jacko Newsgroups: comp.lang.forth Subject: Re: A short history of the stages of development and status of RP's Forth interpreter. Date: Sat, 14 Apr 2012 09:04:00 -0700 (PDT) Organization: http://groups.google.com Lines: 42 Message-ID: <28276169.1301.1334419440853.JavaMail.geo-discussion-forums@vbuo5> References: <0097d6f5-cd50-4231-bf29-a4bfa15cfdaf@z17g2000yqf.googlegroups.com> <2012Apr12.170151@mips.complang.tuwien.ac.at> <91f6aba9-99b7-4dd3-9d30-e19bde3430af@n10g2000yqg.googlegroups.com> <2012Apr13.124255@mips.complang.tuwien.ac.at> <2012Apr13.174811@mips.complang.tuwien.ac.at> <4f88a2cd.157452370@192.168.0.50> <96bd986a-eb9d-451e-8d50-f52d9d319066@x17g2000yqj.googlegroups.com> NNTP-Posting-Host: 92.40.254.35 Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: quoted-printable X-Trace: posting.google.com 1334420219 24242 127.0.0.1 (14 Apr 2012 16:16:59 GMT) X-Complaints-To: groups-abuse@google.com NNTP-Posting-Date: Sat, 14 Apr 2012 16:16:59 +0000 (UTC) In-Reply-To: <96bd986a-eb9d-451e-8d50-f52d9d319066@x17g2000yqj.googlegroups.com> Complaints-To: groups-abuse@google.com Injection-Info: glegroupsg2000goo.googlegroups.com; posting-host=92.40.254.35; posting-account=kZanLQoAAABvNhBbAlX1SsCxeprjdiHJ User-Agent: G2/1.0 Xref: csiph.com comp.lang.forth:11286 On Saturday, 14 April 2012 00:14:08 UTC+1, BruceMcF wrote: > On Apr 13, 6:26=A0pm, "Elizabeth D. Rather" wrote: > > I completely agree with Stephen on this. It is possible for an XT to be > > a direct connection to the executable portion of a word -- that is its > > purpose. A name may be in a different data space, on a different > > computer, or even non-existent. Having it be the primary runtime > > identifier of a word would require some form of indirection, adversely > > affecting performance. >=20 > As I understood it, the NT proposals I have seen is rather to preserve > the XT's role, being as direct a connection to the executable portion > of a word as practicable, and not overload onto it the means of > accessing a name entry in a dictionary, given that a name may be in a > different data space, on a different computer, or even non-existent. >=20 > This was in the face of the alternate proposal to address the > challenges of writing portable programming aids that the XT also be > sufficient to recover information about the dictionary entry that > refers to the executable portion of a word, which requires the XT to > support some form of indirection, adversely affecting performance. I find that in jarvar, my language I'm developing, a lot of indirection is = required to bring a variable onto stack, but the gain is an effective means= of decompiling a word definition. Or it will be when I add a SEE word. > > Moreover, there are reasons to give implementers broad latitude as to > > the construction of name fields: they may include additional informatio= n > > (such as a link to source files, usage counters, all sorts of stuff) or > > be highly compressed in some fashion. There are good reasons in > > different contexts for doing all of these things. If an NT is to be of > > use, however, something must be standardized about what is at that > > destination, constraining implementations. >=20 > The balance of the NT proposals I have seen are words that operate on > NTs, and do not entail standardization of the contents of dictionary > entries at all. They rather refer to accessing information that we > know there exists to know about a dictionary entry, given already > existing standard specifications. So many NT operation words and a find like word to make an NT into an XT, w= hich is what find does. But is not an NT then just a compact string form?