Re: Re: [xml] 2.5.x performance vs. 2.4.24

[Mark Itzcovitz <mark itzcovitz ntlworld com>]

>   err, what did I miss again ?

The following, from Monday:

-------------------------------------------------------
> However I'm unable to reproduce your problem:
>
> paphio:~/XML -> file slashdot16.xml
> slashdot16.xml: Little-endian UTF-16 Unicode English character data 
> paphio:~/XML -> ls -l slashdot16.xml
> -rw-------    1 veillard www         10374 Aug 11 14:07 slashdot16.xml
> paphio:~/XML -> ./xmllint --noout slashdot16.xml paphio:~/XML -> 
> ./xmllint --version
> ./xmllint: using libxml version 20509
>    compiled with: FTP HTTP HTML C14N Catalog DocBook XPath XPointer 
> XInclude MemDebug Unicode Regexps Automata Schemas paphio:~/XML ->

You had me confused for a while - I couldn't get it to go wrong either. Then I realised I was looking too 
closely at how you'd tried to reproduce the problem. It only occurs outputting the xml, so it doesn't happen 
if you put "--noout" on the command :-). 

>  I had to fix xmlParseEncodingDecl() because "utf-16" is not 
> registered directly as an encoding (only le and be variant are, maybe 
> this should be fixed). But the converters seems to work for this case. 
> This does not mean your patch is wrong, just that I can't reproduce 
> the problem, and I would feel better if I could reproduce it.
>  I didn't realized that iconv would take over libxml2 internal UTF-16 
> parsing when present, seems I would need to do regression testing with 
> multiple configuration flags to actually catch up some of the 
> problems.

Your need to fix xmlParseEncodingDecl() seems to be as a result of a change in 2.5.9 - it works for me with a 
2.5.8 xmllint but not 2.5.9.
-------------------------------------------------------

Also Alex Dron's posting from Tuesday evening is relevant, and my reply.

-----------------------------------------
Email provided by http://www.ntlhome.com/