W3C home > Mailing lists > Public > public-xml-processing-model-wg@w3.org > June 2007

Re: can we have last() having a consistent value ?

From: Norman Walsh <ndw@nwalsh.com>
Date: Wed, 06 Jun 2007 09:31:34 -0400
To: public-xml-processing-model-wg@w3.org
Message-ID: <87abvdgqfd.fsf@nwalsh.com>
/ ht@inf.ed.ac.uk (Henry S. Thompson) was heard to say:
|> 2. Set last() = MAXINT and explain to users that that's the way
|>    it works in V1.
| This avoids the above-noted problem with (1).

It does. I'm uneasy about it because it's always a lie. OTOH,
position()=last() is almost always a lie so I could live with it :-)

|> 3. If the pipeline author uses last() then the step has to buffer
|>    and give the right answer. If that causes a resource error in the
|>    implementation (out of memory, for example), so be it.
| As stated before, I _really_ don't want implementations to have to
| peer into XPaths, so I _really_ don't like this.

I don't want implementations to have to do any real analysis, but
"indexOf(xpath,'last()') >= 0" doesn't seem overly burdensome.

I expect implementors will do some analysis anyway. My implementation
attempts to stream p:delete operations, for example, by considering
the XPath.

                                        Be seeing you,

Norman Walsh <ndw@nwalsh.com> | There is nothing which human courage
http://nwalsh.com/            | will not undertake, and little that
                              | human patience will not endure.--Dr.
                              | Johnson

Received on Wednesday, 6 June 2007 13:32:30 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 21:32:43 UTC