ID attributes in cache files

Cache files define section IDs to contain the page ID, as you would link
to it, such as:

<section id="the_page_id#the_section_id">

This is not how section IDs are written in pages, of course. This is
designed to make it easier to use xsl:key to look up the target node in
the cache file.

Problem: Mallard 1.0 defines the id attribute to be NMTOKEN, which does
not allow the # character. Because you're not supposed to put the #
character in the id attribute in actual pages.

The cache schema is designed to be mixed with the Mallard schema using
the version attribute, such as:

<cache version="cache/1.0 1.0">

Right now, with the cache schema, it is impossible to validate because
of this mismatch. There's no easy to to override the type of the id
attribute in the 1.0 schema. It's just not designed to allow that to be
overridden. We can't change the 1.0 schema now. But we can change the
cache/1.0 schema.

Attached patch fixes the validation issue. But it's ugly and fragile and
could potentially lock out extensions from validating in cache files. It
has to be done, or cache/1.0 will never validate. But we need to look at
how to make the 1.1 schema more extensible so cache/1.1 can drop this
awfulness.

--
Shaun

-------------- next part --------------
A non-text attachment was scrubbed...
Name: cache-id.diff
Type: text/x-patch
Size: 1357 bytes
Desc: not available
URL: <http://projectmallard.org/pipermail/mallard-list/attachments/20150710/4434a850/attachment.bin>