- From: Andrew Layman <andrewl@microsoft.com>
- Date: Wed, 21 May 1997 14:00:37 -0700
- To: "'Steven J. DeRose'" <sjd@eps.inso.com>, w3c-sgml-wg@w3.org
I just asked several parser writers about the implementation difficulty of single versus double colons. They said that single are easier and will lead to simpler, smaller parsers. > -----Original Message----- > From: Steven J. DeRose [SMTP:sjd@eps.inso.com] > Sent: Wednesday, May 21, 1997 8:15 AM > To: w3c-sgml-wg@w3.org > Subject: Re: SD5 - Namespaces > > At 09:22 AM 05/21/97 +0100, Peter Flynn wrote: > >Murray Altheim writes: > >> I can't remember who mentioned it initially, but I kinda prefer two > >> colons "::" to one ":", since it mirrors FPIs' namespace delimiter. > >> Also, it's more unique (and hence explicit) than a single colon. > >> Either way, agreed: a neat idea. > > > >I agree. This presumably just means a change to the SGML Dec to allow > >the colon in names? > > True; but remember that we cannot express in SGML the claim that ":" > has a > special meaning within GIs. > > We could do so in XML, of course (though it would be tricky to allow > single-colon as any old namechar, yet reserve double colon as a > delimiter). > > gi ::= gipart > | gipart "::" gi > > gipart ::= namestart namechar* > > Steven J. DeRose, Ph.D., Chief Scientist > Inso Electronic Publishing Solutions > (formerly EBT)
Received on Wednesday, 21 May 1997 17:00:47 UTC