Re: [xliff] csprd01 014 MetaData Module lacks processing requirement

Subject: Re: [xliff] csprd01 014 MetaData Module lacks processing requirement

On Mon, Jul 1, 2013 at 9:44 PM, Schnabel, Bryan S <bryan.s.schnabel@tektronix.com> wrote:

I propose we do the following to enable the capture of attributes for an XML roundtrip, using the maximalist method (i.e., to not use a skeleton):

(1) make the processing requirement: “agents MAY ignore the <metadata> element, but the MUST NOT remove it.”

Since this is one of our general processing requirements, we do not want it repeated with every module. Modules MAY be ignored, because they are optional, and MUST NOT be removed, because they are XLIFF defined and you MAY rely on them for roundtrip (if you can deal with the fact that they MAY be ignored)..

If this is not clear then we need to make this clear in the core/general location, not add more or less consistent repetitions in every module.

(2) Change 2.1 form "Should" to "MUST": "A tool processing a valid XLIFF document that contains custom elements that it cannot handle MUST preserve those elements."

I do not think we want to change this, there was a ballot that decided that modules deserve absolute protection (MUST), while custom (i.e. namespaced extension elements) SHOULD be preserved, since there well may be valid reasons for deleting them, they can e.g. compete with a core or module feature, be too big, in conflict with another extension you prefer etc.

We may want to clarify that custom elements means (namespace based extension elements other than XLIFF defined modules.

mda does not have any custom elements IMHO, it only has user defined values.

Note: 2.7.2 seems to imply that the maximalist method must not be used in an XML roundtrip by prohibiting the use of the Metadata module for creating a translated version of the original document. The TC has not agreed to cease support for the maximalist method, so I propose the following:

(3) Change 2.7.2 form "must" to "SHOULD"; and limit the clause to just extensibility: change

"Tools must not rely on user extensions (either in the Metadata module or custim [typo] namespace based) other than the ones possibly defined in <skeleton> to create the translated version of the original document." to "Tools should not rely on custom namespace based extensions other than the ones possibly defined in <skeleton> to create the translated version of the original document."

I do agree that we should not prohibit use of mda carried metadata for roundtripping, since their survival is guaranteed by the general PR.

However, relying on custom extension should still be absolutely prohibited (MUST NOT). "SHOULD" would be opening a backdoor for extensibility misuse and undermine interoperabilty in the XLIFF 2.0 eco-system. Our core contains everything that is essential for a successful L10n roundtrip. modules are protected, it is fair to say that you MUST not rely, on extensions, since these are NOT absolutely protected.

Note: I am purposefully not addressing the use of the Metadata module as it impact the discussion on re-segmentation. I’ll leave that in the other thread.

Please let me know if you have any questions or objections by July 8. If not I will consider this matter resolved and make the updates to the spec.

Thanks,

Bryan

xliff message