Re: How Do I Generate A Set-Difference With Context -

On Sun, 13 Mar 2005 00:52:23 +0000, Aron Bock <aronbock@xxxxxxxxxxx> wrote:

> Now for a question: each <story> has, say, a @date attribute; multiple
> stories in 1 feed may have the same @date value.  If I find a changed (or
> new) story, I want to pass through not just that story, but also all other
> stories with that same @date value, no matter they're unchanged.
> 
> My first thougth about accomplishing this is to perform some sort of
> pipeline processing (it's immaterial if it's done in-situ in the XSLT
> processor).  The first pass would generate a diff result, and a second pass
> would add back all stories of the same [unique] dates as the ones in the
> diff result.
> 
> It would be nice to do this in one [conceptual] pass:
> --for every story check if it should pass through
> --if a passthrough also pass through otheries stories with the same @date
> 
> Of course, a naive approach would have stories output multiply, unless I
> somehow "remember" what [say, dates] has already passed through.  I could do
> that in a procedural language, but don't know how to [efficiently] in a
> functional one such as XSLT.  While I read up on functional programming
> techniques, if anybody can suggest a way to accomplish this, please post.
> 
> Here are 2 sample feeds:
> 
> feed1.xml
> ======
> <feed>
>  <story @date="1" text="a"/>
>  <story @date="2" text="a"/>
> </feed>
> 
> feed2.xml
> ======
> <feed>
>  <story @date="1" text="a"/>
>  <story @date="1" text="b"/>
>  <story @date="2" text="a"/>
> </feed>
> 
> If feed2.xml were the "current" feed, and feed1.xml the propr one, I want to
> end up with the following, in 1 pass.
> 
> final.xml
> =====
> <feed>
>  <story @date="1" text="a"/>
>  <story @date="1" text="aa"/>
> </feed>
> 
> Thank you
> 
> --A

This is confusing:

   There is no "text" attribute with value "aa" in the two xml documents.


It's true that nobody's replied to the first post until now. I didn't,
because I didn't find a complete and clear description of the problem
-- one has to put some effort in specifying a problem in precise
manner, this is a logical pre-condition for people being able to help.

Additionally, if performance is to be optimised, this requires that
real (or at least realistic) input files are included.

Cheers,
Dimitre Novatchev.