[Koha-devel] State of the debian packaging

[Vincent Danjean]

Frederic Demians wrote:
>>> And if we know where the XML parser is called (directly or indirectly),
>>> we can set in the koha file(s) the global variable
>>>>> $XML::SAX::ParserPackage = "XML::LibXML::SAX";
>>> in order to force the good XML parser without requiring to modify a
>>> global configuration file
>> Primarily, Koha uses a SAX parser in MARC::File::XML, but also 
>> XML::Simple.
> 
> XML::LibXML::SAX is specifically required by MARC::File::XML. With 
> another SAX parser, XML MARC records are not correctly decoded-encoded 
> and accented characters, for example, are mixed.
> 
> In this file:
> 
>     MARC/File/XML.pm
> 
> line 21:
> 
>     my $factory = XML::SAX::ParserFactory->new();
> 
> a SAX parser is instantiated.
> 
> So since MARC::File::XML need XML::LibXML::SAX, it should ask explicitly 
> this parser and insert before line 21, as said by Vincent:
> 
>     $XML::SAX::ParserPackage = "XML::LibXML::SAX";
> 
> Can't someone contact MARC::File:XML package maintainer and ask him to 
> update his code?

Someone can ask but I'm not sure that the MARC::File:XML package maintainer
will agree with this modification. Perhaps, he will add the line. But he
can also answer:
- MARC::File::XML can work with other XML parser if there is no accentuated
  letter (no non-ASCII letter) so I do not want to force a specific dependency
or
- MARC::File::XML requires a correct XML parser. If one XML parser is buggy
  then bug must be filled against this XML parser and not against packages
  that use XML parser.
or
- MARC::File::XML will implement in a future release a mechanism to be able
  to choose the XML parser (such as XML::Simple)
or
- ...

So, asking him is a good thing (perhaps he does not know the issue). But
koha can also force the use of the parser it needs (ie add
$XML::SAX::ParserPackage = "XML::LibXML::SAX";
when it uses MARC::File::XML)

It would be good also that misc/sax_parser_test.pl really tests the parser
(I get the same binary output with XML::LibXML::SAX or with the Expat XML
parser)

Note that, according to the XML::Simple documentation, XML::Simple can also
force the XML parser it will use by setting:
$XML::Simple::PREFERRED_PARSER="XML::LibXML::SAX";

I prepared two commits for koha to force the use of the LibXML parser. Can you
test them and apply them if they are ok ? (I only greped for any occurrences
of XML::Simple and MARC::File::XML and add the correct line).
Both commits are available at the top of my wip-for-upstream branch:
http://git.debian.org/?p=collab-maint/koha.git;a=shortlog;h=refs/heads/wip-for-upstream
They are:
- "Force MARC::File::XML to use the XML::LibXML::SAX parser"
and
- "Force XML::Simple to use to XML::LibXML::SAX parser"

  Regards,
    Vincent