RE: SGML Unique IDs

[edunn@xxxxxxxxxxxxxxxxxxxxxxxx]

Thank you for the replies to date.

Bill Hall mentioned the advantages of document management. We do in fact have a
SGML validation/submittal and document management system for our suppliers (for
submittal, validation, approval, and versioning of fragments) coupled with
Astoria for final document assembly and content management. Unfortunately, the
documents I'm having problems with predate the system.

Chris Despopoulos understands my problems exactly. Except my mention of a book
file was only for the purposes of outlining the script I was thinking of
writing. Currently we have a script that prefixes the file name to certain IDs
(PERL on MIF, although as John Root mentioned I suppose VB or VBA could work
too) and I think our quick fix will be to have that script expand to all ID and
RefID attributes.

The duplicates appear in different documents. As Lynne Price pointed out, you
can't have duplicate IDs in a file. I just wish FM had a built in manager for
IDs on a book level. Once the files are all assembled into a book you can
validate for duplicate IDs but there's no easy way to correct the mess.

All of our problems are from users reusing SGML documents as templates instead
of starting new docs or copying fragments from one doc to another. This time
round there are no external xrefs, so the fix can be rather banal. I just hope
that our Doc management system is implemented in such a way we can properly
follow/maintain inter document links in the future.

I was trying to find a way to appear smarter I suppose. Creating a temporary
file and creating a xref to have FM create and ID then using that ID to replace
an existing one came to mind. Also moving away from MIF would be a huge
advantage as it takes an eternity right now to process the files (1000s of em
and few of them are small). Round tripping to SGML and doing PERL on the sgm
files would be faster but that mangles many things that we need for the paper
output that we don't need in the IEM. FDK or FrameScript should be very fast.

Will have to see what we have the time and resources to accomplish.

Eric L. Dunn



** To unsubscribe, send a message to majordomo@omsys.com **
** with "unsubscribe framers" (no quotes) in the body.   **