ID values in inline code and XLIFF in general

[Yves Savourel <ysavourel@enlaso.com>]

Hello one and all,

As we were making progress on sub-flows in the Inline Markup SC, we reached a point where we have an attribute (called 'subFlows' for now) that can store a list of IDs.

The idea is that that attribute points to the different units where each sub-flow is stored.

For example, if you have this original HTML:

"Some text with <img title='Title text' alt='Alt text' src='image.png' />"

We would have this type of XLIFF:

<unit id='100'>
 <segment>
  <source>Title text</source>
 </segment>
</unit>
<unit id='200'>
 <segment>
  <source>Alt text</source>
 </segment>
</unit>
<unit id='300'>
 <segment>
  <source>Some text with <ph id='1' subFlows='100 200'/></source>
 </segment>
</unit>

where the subFlows attribute lists the IDs of the two <unit> elements where the title and alt texts are.

The question we have is:

What characters should be allowed in an ID in XLIFF?
We need to know because the syntax of the subFlows value may be different depending on that (namely: it may change the separator between IDs).

A few notes:

a) The xsd:Id type is probably out because it forces a unique value per document, and most our IDs can be multiple within the document (while some are being unique in their context).

b) While we cannot think of everything, it may be wise to think about common IDs and allowance in other systems (DBs, UUIDs, Resource files, XPath, etc.). So we don't block out much.

c) For information, an ID can include just about anything in 1.2 (including space).

Any thoughts?
-yves