Re: [xsl] XSLT, Web applications, and "native XML databases"

Subject: Re: [xsl] XSLT, Web applications, and "native XML databases"
From: "Andrew Welch" <andrew.j.welch@xxxxxxxxx>
Date: Thu, 8 Jan 2009 10:09:09 +0000
>>>> Andrew Welch wrote:
>>>>> Once XSLT can work directly directly against
>>>>> the database, XQuery's days could be numbered.
>>>> Bit sweeping Andrew, but it always appeared as
>>>> a fairly natural extension to XSLT, to have
>>>> it operate on databases rather than files, to keep the dataheads
>>>> happy and answer the users who want to play
>>>> with dirty great datasets.
>>>>  I'm surprised it's not on the agenda for the WG
>>> If you really want XSLT to work directly against the database
>>> you could use collection() to return the documents.
>> false.
> How about URIs (e.g. with protocol) directly referencing eXist collections
> with i/o "as usual" for XSLT (e.g. doc(), document(), collection(),
> result-document()) ?

To explain the difference between what is available now, and what we
mean by XSLT 2.0 working directly against the database, think of this

- You have 1000 xml files stored in the database, and each of them
have a <title>
- You want to get those titles and group, sort and number them using xslt 2.0
- Currently you either a) use xquery to get the title nodes, wrap them
in an arbitrary root element, then supply that document to the
transform (which needs to know about that new document) using the
database's xslt extension, or b) access all documents using say the
rest api and then extract the <title> from within the xslt...  not
very feasible, and not taking advantage of the database.
- When xslt can work directly on the database, you would just do
<xsl:for-each-group select="collection(...)//title">

Andrew Welch

Current Thread