[Prev][Next][Index][Thread]

Re: ERB: decision and conundrum



> 1. Contrary to our decision of last time, we will support subelement
>    addressing by a simple search operator.  We will make it clear that
>    bit-for-bit matching without respect to words or tokens is compliant
>    behavior; if implementations wish to compete on the basis of 
>    case-folding or other fancy search optimization, that's fine.


If I understand this, you're saying that that a simple string search
(sensitive to changes in whitespace, e.g. the CR/LF substitution that
that seems to happen to text MIME media types over HTTP) will be used,
but if it isn't found, an implementation can find something else
instead -- e.g. using case substitution, or maybe a thesaurus --
and claim a match??

Can I use POSIX regular expressions?

> 2. Locators shall consist of a URL, optionally followed by a '#'
OK, good.

> 3. The '#' may be followed by the string "<tei>", in which case the
>    remainder of the locator is to be treated as a TEI extended pointer.

Fine.

[...]

> On another subject, we agonized further over the fact that current
> implementations of '#' in URLs always fetch the whole document and
> then navigate to the fragment in the client.

This isn't correct for Netscape on Unix at least -- if the entire
document is already loaded, it doesn't fetch it again.  This depends
on your options, of course, and the Expires: header -- if the document
is expired, hor has a pragma: nocache associated with it, it will
be fetched again.  But it isn't usually.

Or do you mean that you want them not to fetch anything prior to
the given ID in the document??? Eeek, that's no good!  That's an
entirely different sort of thing!  Dynamic fragmentation can be
done with other mechanisms.

> 4. If the '#' is followed only by a string, then.... what?

Look to see if /usr/games/bin/rogue is installed and fire it up.
[old gcc ANSI standards compliance joke]

Seriously, if you only allow TEI pointer, then you don't need <tei>.
If you allow other things, what are they?

How about
    #<att>sec1.2
to search for any attribute equal to sec1.2?

Anything more complex must be done with TEI ptrs.

I also note that you have no space for CGI queries or for MIME
parameters.

If they are present, where do they go?

    "http://xxxx/yyy/zzz#xxx;version=3.2"
or
    "http://xxxx/yyy/zzz;version=3.2#xxx"
?

What about
    "http://xxxx/yyy/zzz&myname=liam&myfeet=bare;version=3.2#xxx"
?

Finally, probably need to note the need to use ; instead of & here
in XML CDATA attributes, since CDATA means RCDATA for attributes, in
order to make the SGML standard harder to understand.
    "http://xxxx/yyy/zzz;myname=liam;myfeet=bare;version=3.2#xxx"
or
    "http://xxxx/yyy/zzz;myname=liam;myfeet=bare#xxx;version=3.2"
?

Lee