Re: [xslt] ready to use extensions??

[Janning Vygen <vygen gmx de>]

Am Mittwoch, 24. April 2002 14:10 schrieben Sie:
> On Wed, Apr 24, 2002 at 01:45:12PM +0200, Janning Vygen wrote:
> > Hi,
> >
> > are there any ready to use extensions for libxslt (i know that
> > you cant link dynamically, so i mean ready to compile source code
> > :-)
>
>   well everything in the libexslt subtree of the libxslt source
> are basically extension functions.
>   I also suggest you read the tutorial on the subject:
>      http://xmlsoft.org/XSLT/extensions.html

i did already of course :-) 

> > I would like to have an extension which can get imagewidth and
> > imageheight of a referenced image, so i need filesystem access
> > for this, which can only be done with an extension.

but maybe still anybody has written such an extension??? I guess a 
filesystem extension is needed very often (download size, imagesize, 
file available, url available, newest source in directory and so on.)

Ok, its not portable anymore as you mentioned, but i can use the 
function-available() feature before to keep it portable.

good examples for the need of those extensions: 
if you dont use unparsed entities for an image reference, but only 
href attribute in docbook you cant say if this image is available It 
would be nice to produce an error if its not

How about a link list. it would be fine to check the url attribute of 
an ulink in docbook and produce an error or just render the link a 
different way if its broken.

Is this the right list for asking?? How many subscribers are here? 
Maybe i should query the mulberry xslt list??

> > btw: i guess the extension handling is one of the drawbacks of
> > libxslt[1]. in saxon you can call Java API from inside an XPATH
> > Expression. So you can for example get the filesize of a
> > referenced file and put this in the result tree like "Download
> > now (500 KByte)"
>
>   Well, but it's an extension to the specification. I.e. it's not
> portable anymre.
>
>   again 1.1 is dead and automatic conversion to node-set is a
> relatively strong violation of the specification. This will be
> relaxed in 2.0 if I understand correctly, so at some point this
> will get fixed.

i know 1.1 is dead, but how to call that stuff saxon does?? Most 
people know this as 1.1 because in many books its called like this.

thanks,
janning

-- 
Planwerk 6 /websolutions
Herzogstraße 86
40215 Düsseldorf

fon 0211-6015919
fax 0211-6015917
http://www.planwerk6.de