Advancing the GCT Proposal

Earlier this year, I had the opportunity to spend an afternoon with Robin LaFontaine discussion document formats and, specifically, the GCT Proposal, which has been offered here as the basis for a Community Specification.

I am offering my full support for the advancement of the GCT Proposal.  That is predicated on the following assumptions:

 1. The GCT Proposal is a general mechanism for specifying the modifications by which one XML Document is produced by identified transformations made to another XML Document.

 2. The pre- and post-transformation documents are valid XML documents in whatever sense "validity" is assessed, whether via explicit DTD, appeal to a schema, etc.  This has to be explicitly established.

 3. It is assumed that, for any kind of semantic determination, the preservation of that in GCT is established by other means.

I have two concerns.  These might already be addressed in GCT.  I need to check them out.

 A. The first concern is about the reversibility of changes in terms of what must be reversed atomically.  That needs to be dealt with.  One thought is that the only reversals that are legitimate are ones for which the resulting (intermediate) document is also valid the same way as reversing all the changes restores an original starting version.

 B. My other concern is when reversal needs to be just another change.  That is, there is no loss of manipulation history, and any metadata about a change, when reversing that change.  Sometimes it is important to eradicate a (part of) revision history.  But it would be ideal if there was a way to also specify reversal of a change via another recorded change, so there is an account of such revision activity.  (I will not worry too much about reversing a reversal, which now becomes possible.)

I believe there can be support for semantic issues, but I think that should be done as supplementary work. I think GCT can carry markers for semantic constraints and the related semantic profiles, but it need not do anything but provide for such markers.

I think that should be a separate and interesting discussion.


 -- Dennis E. Hamilton
    orcmid@apache.org
    dennis.hamilton@acm.org    +1-206-779-9430
    https://keybase.io/orcmid  PGP F96E 89FF D456 628A
    X.509 certs used and requested for signed e-mail

Received on Monday, 15 June 2015 17:24:10 UTC