Re: [xml] Canonical XML Implementation

[Daniel Veillard <veillard redhat com>]

On Sun, Mar 03, 2002 at 11:41:08AM -0800, Aleksey Sanin wrote:
> Daniel,
>
> Please find my answers interlaced with your questions.
>
> Aleksey.
>
> > I fetched the tarball, I assume it's the same as would have been in the
> > attachment ;-) . Did you add a configure option to include/exclude it in/from
> > libxml2 ?
> Both tarballs have only files for C14N. The attachment I sent has only 
> *.h and *.c files.

  Well actually I didn't got it. that's why I asked :-) I will double check

> The tarball on the Web site has also Makefiles, test program and test suite.
> I have not enought expirience with autoconfig scripts so it'll be great 
> if you can take care
> of adding include/exclude option in libxml2. I simply not sure that I 
> can do it right.

 Sure.
  
> > The code looks good, I will need to ask the Gnome2 release team for the
> > integration at that point but I doubt it will be a problem.
> Great! Please let me know if any changes will be required.

  I don't think so.

> >     - are the functions xmlC14NProcessNode, xmlC14NProcessNodeList
> >       and xmlC11NNormalizeString really part of the API ? What would use
> >       it, XML-DSig for example ? I would be tempted to not expose the
> >       content of xmlC14NCtxPtr and keep it defined in c14n.c (but keep
> >       the pointer definition public), any reason this would not work ?
> You are right, there is no no need in these functions. Whole 
> c14nInternals.h file could be placed
> at the beginning of the c14n.c w/o any problems. Do you want me to do 
> this change?

  I will take care of it. I'm still wondering about keeping some semi-public
entry points for those..

> >     - The regression tests from the tarfile should really be integrated
> >       as part of libxml2 regression tests, the simplest seems to add 
> >       a --c14n option to xmllint.
> It does not look simple since there are 4 different options 
> (--c14n-with-comments,
> --c14n-without-comments, --exc-c14n-with-comments and 
> --exc-c14n-without-comments)
> and all these options accept different sets of input parameters. 
> However, I'll take a look.

  Okay I see. some automated regression testing are needed anyway. I was
also wondering if a function doing the attribute completion from DTD defaulting
would be useful in case the tree wasn't parsed with the expected flags.
Can probably be postponed a bit.

> > Since "Exclusive XML Canonicalization" is in CR, it would be very useful
> > to the Working Group if you could send them implementation feedback (authors
> > and w3c-ietf-xmldsig w3 org  This will definitely help them making progress
> > toward reaching Recommendation status.
> Yep, I already sent few questions to the Working Group.  I am also going 
> to send them
> information about C14N implementation in libxml2 (when it'll be included 
> in the package)
> if there is no objections from your side. There is an interoperability 
> reports pages and I think
> that it's a good idea to list libxml2 on these pages

  Sounds excellent !

Daniel

-- 
Daniel Veillard      | Red Hat Network https://rhn.redhat.com/
veillard redhat com  | libxml Gnome XML XSLT toolkit  http://xmlsoft.org/
http://veillard.com/ | Rpmfind RPM search engine http://rpmfind.net/