Re: Revision Marks (was) Re: Commit (HEAD): SDW: Metadata

From: David Chart (linux@dchart.demon.co.uk)
Date: Fri May 17 2002 - 16:51:57 EDT

  • Next message: David Chart: "Re: Revision Marks (was) Re: Commit (HEAD): SDW: Metadata"

    On Fri, 2002-05-17 at 20:44, Tomas Frydrych wrote:
    > > This doesn't strike me as particularly user-friendly.
    >
    > It is exactly how Word works, so I expect that is what the users
    > would expect / put up with. But then why not do it better.
    >

    Exactly. No need to just play catch-up. :-)

    > > I'd like a revision-stet attribute, which you add when you say that you don't
    > > like a particular revision. If that revision level is active, then the
    > > revision is nixed, although you should be able to see that it was
    > > made.
    >
    > That will only work up to 3 changes to the same text, but we could
    > allow the revision-remove and revision-add attributes to carry
    > multiple revision ID's, and this way we could achieve an infinite
    > depth of such changes even without the revision-stet attribute.

    Can we nest attributes? We are going to have to introduce properties to
    hold the attributes anyway, as people are not always going to revise
    things that already have <c></c> round them. (I've changed single
    letters in the middle of a word before now.)
     
    > What I am not sure about is how to indicate in the text, in a
    > sensible way, the number of actual changes associate with a
    > particular piece of text. The advantage of my previous suggestion
    > was that you never had a piece of text belonging to more than two
    > revisions, but I am sure we can come up with something.

    That's not an advantage, I think. Suppose 1 deletes a sentence, 2
    retypes, 3 deletes, and 4 retypes. Supreme Command agrees with 2 & 4,
    and accepts their revisions. You now have three copies of the sentence.
    (The original, 2's retype, and 4's retype.) If we're going to allow
    acceptance of revisions by reviser (which would be cool), we need to
    keep track of what the revision applies to.

    Actually, I think the logic here needs some serious thought before any
    coding is done. It should be easy, if we design the initial framework
    well. Since this doesn't actually involve writing code, I can help, I
    think. Although possibly not right now -- I need to sleep.

    -- 
    David Chart
    


    This archive was generated by hypermail 2.1.4 : Fri May 17 2002 - 16:55:59 EDT