This is an archived snapshot of W3C's public bugzilla bug tracker, decommissioned in April 2019. Please see the home page for more details.
The XSLT transformation for chameleon schema documents appears not to handle some attributes that can contain QNames. It handles any attribute typed as a QName, and it handles the memberTypes attribute of xs:union (which is a list of QNames), but it does not handle: xs:element/@substitutionGroup (also a list of QNames) xs:any/@notQName (a list that can include QNames inter alia) It is presumably by design that it does not tamper with XPath expressions, which may also contain unprefixed QNames.
WG agrees, marking as needsDrafting since MSM may have further input.
Created attachment 1057 [details] non-schema-aware version of chameleon include transformation I believe it doesn't tamper with XPath expressions because the WG was persuaded that the right thing to do was to specify that XPath expressions in a chameleon-included document could and should be interpreted with the new target namespace as the default namespace. I attach a version of the chameleon include stylesheet (not yet tested) which handles all the attributes declared in the schema for schemas as being or including QNames.
Created attachment 1058 [details] revised schema-oblivious stylesheet I've now tested the revised stylesheet on a simple schema document and have corrected a couple of errors in the earlier attachment.
I have compared this carefully with the version I have been using internally in my implementation. Generally it looks OK and better than my attempt. I think that the only fault I can find is that it would be better to replace resolve-QName(string(.), ..) with resolve-QName(normalize-space(.), ..) on lines 34 and 38, because resolve-QName (probably) does not strip leading and trailing whitespace from its argument. (One might also common up this expression into a variable.)
The revised transform attached on 29 December (see comment 3) amounts, I guess, to a wording proposal that would resolve this issue. So I'm marking this bug as needs-review.
At its call today the XML Schema WG adopted the proposal to replace the chameleon-include transformation with the version given in an attachment to this issue. That has now been done and the new version is appearing in the status-quo document, so I'm marking the issue resolved. Michael, as originator would you check to make sure the change has really been made, and then close the bug? Thank you.