- From: Felix Sasaki <fsasaki@w3.org>
- Date: Wed, 01 Feb 2006 12:06:51 +0900
- To: "Mark Baker" <distobj@acm.org>
- Cc: public-cdf@w3.org, public-i18n-core@w3.org
Hello Mark, thanks for your reply. On Tue, 31 Jan 2006 10:04:21 +0900, Mark Baker <distobj@acm.org> wrote: > > Hello Felix, thanks for your comments. > > On 1/25/06, fsasaki@w3.org <fsasaki@w3.org> wrote: >> >> Comment from the i18n review of: >> http://www.w3.org/TR/2005/WD-CDR-20051219/ >> >> Comment 5 >> At http://www.w3.org/International/reviews/0601-cdf/ >> Editorial/substantive: S >> Location in reviewed document: >> general >> >> Comment: >> >> If you ask an SVG document about language information, and the document >> is inside an HTML document, the xml:lang attribute in the HTML applies >> to the SVG as well. It seems that the compounding specs should say: >> \"You should get the same results for both inclusion and referencel.\" > > The WG has just discussed this, and we feel that for the CDR case - > which is all the current set of Last Call drafts cover - the value of > the xml:lang attribute in any containing HTML should *not* apply to > children, because it isn't authoritative (as described in the TAG's > finding on authoritative metadata[1]) as a result of requiring > multiple messages to assemble the compound document. Consider, for > example, that the child document might be returned with an HTTP > message which includes a Content-Language header (sec 14.12 of RFC > 2616) with a (authoritative) value inconsistent with that specified by > the xml:lang attribute. More generally too, content may be retrieved > from multiple domains over which the author of the containing document > has no control, and therefore propagating the value of attributes like > xml:lang doesn't seem appropriate. > > [1] http://www.w3.org/2001/tag/doc/mime-respect.html > > Thanks. > > Mark. > The i18n core working group still disagrees with your reply. We have a scenario in mind with an XHTML document which contains an SVG image. If it is external, the language information does not apply anymore. Our concern is: what should happen if there is no "authoritative metadata"? We agree with the explanation you gave us, but would like to have a scenario for the case without "authoritative metadata", but with document internal data. Regards, Felix.
Received on Wednesday, 1 February 2006 03:07:11 UTC