"Re: [xslt] Apply stylesheet to a subbranch of a document"



Daniel Veillard wrote:

> On Tue, Oct 07, 2003 at 01:38:52PM +0200, Kasimier Buchcik wrote:

>>>Daniel Veillard wrote:

>>>> Well try the function modification, we will put a big warning in the API
>>>>description
>>
>>This mail refers to this initial mail:
>>   http://mail.gnome.org/archives/xslt/2003-September/msg00019.html
>>
>>We modified xsltApplyStylesheetInternal to provide the ability to apply 
>>a stylesheet to a subbranch of a document. It worked well by simply 
>>changing the initial context node. But we ended up wondering why the 
>>heck we should mess with the specs. Originally I was influenced by the 
>>behaviour of the m$xml 4 parser on that issue. IOW: keep it as it is; we 
>>will find an other way.
> 
>   Hum, okay. I wonder a bit what made you change your mind,
> but that's fine :-)

Let's say some people that I get my money from did not want to take the 
risk to use functionallity that is not in the specs and thus might be 
not available in other parsers. Well, they convinced my at last.

>>Anyway, thanks for the discussion!
> 
>   no problem.
> 
>>By the way: serializing documents took long time if using functions like 
>>xmlDocDumpFormatMemoryEnc, until we discovered that serializing directly 
>>to file was way faster (by the factor of 50 sometimes). I know that 
>>there was a message about this issue already somewhere on the list; so 
>>this is just a hint for those who transform huge documents and process 
>>further with the resulting xml-text. We did a workaround by serializing 
>>to file and then reloading the document.


Actually it's not a libxslt but a libxml2 issue, so you want to move it 
there.

Using libxml2 2.5.10 (Igor's win binaries) I get the following times for 
  the serialization of a 7.7 MB xml-file:

xmlSaveFormatFileEnc      : 2.1 s
xmlDocDumpFormatMemoryEnc : 77.2 s

 > Hum, wasn't that fixed in the last versions ? If not open a bug in 
bugzilla
 > so it get checked and fixed if it's not already.

I don't know if you still confirm that this is an unexpected behaviour. 
Should I still report it to bugzilla?

Thanks,

Kasimier Buchcik




[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]