Re: [xml] minor compile warnings

[Daniel Veillard <veillard redhat com>]

On Wed, Feb 13, 2008 at 12:28:42AM +0200, Roumen Petrov wrote:
> Hi Hannes,
> did patch sent in a previous email work for you ?

  Silence assumed as agreement, patch applied and commited to SVN,
thanks a lot !

> > > If one day libxml2 go international and decide to use gettext system for 
> > > message translation existing check for iconv can be replaced with macro 
> > > AM_ICONV.
> > >    
> >
> >  Well libxml2 is internationalized, it's not localized, and it's 
> >  on-purpose.
> > If you really build an end user application one should catch the error 
> > using
> > the APIs and provide appropriate user feedback, just outputting libxml2 
> > low level errors is in a majority of case not the right thing to do, 
> > logging
> > them for more in-depth analysis is appropriate though. Also i want to 
> > minimize
> > the platform requirement, and even gettext can be a problem for some users.
> >  
> May be we think about different things. I mean i18n capable application 
> that can be localized (l10n).

  Yes understood, and my point is that libxml2 is fully internationalized,
since it can process data using the full unicode set, but not localized.

> I don't see where libintl.h is included and where functions like 
> gettext(...) is invoked.
> I know that libxml2 can work with different charsets (codeset) and can 
> use internal ISO codesets if OS don't support them.
>
> If one day libxml project decide to use localized messages ( as example 
> output from xmllint --help )
> I could assist in integration of gettext system into project.
> Also *internationalisation and localisation is out of scope of this mail 
> thread.

  Well really as explained I think in this kind of uses we are better off
without gettext and localization, I maintain other pieces which uses
gettext, so I have some insight on the trade-off, I don't think the user base
would gain much, libxml2 being buried at the bottom of the application stack,
if libxml2 error messages shows up directly, to me it's an application bug,
they should intercept and process errors. The way it's reported and
processed is really application dependant. The excruciating details of why
something failed to load or validate are in a large majority of XML processing
case of no interest to the end user (but crucial to the app developpers and
sysadmins who I guess can handle english).
  xmllint is a bit special in this respect, I agree but i prefer to not
add the dependancy to gettext just for that command line application.

Daniel

-- 
Red Hat Virtualization group http://redhat.com/virtualization/
Daniel Veillard      | virtualization library  http://libvirt.org/
veillard redhat com  | libxml GNOME XML XSLT toolkit  http://xmlsoft.org/
http://veillard.com/ | Rpmfind RPM search engine  http://rpmfind.net/