Re: IMPORTANT: proposed removal of non-bidi code

From: Paul Rohr (
Date: Thu May 09 2002 - 19:28:24 EDT

  • Next message: Andrew Dunbar: "Re: newer press release"

    At 09:17 PM 5/9/02 +0100, Tomas Frydrych wrote:
    >The other thing is, that with the rewrite Martin has started I suspect
    >we will not have a properly working code in the HEAD for some
    >time -- my understanding was that this was precisely the reason for
    >the 1.0 release, so that we could do radical things to the head.

    Gosh, that's not the sense I've gotten.

    Moving HEAD on to a new unstable tree allows people to start doing much more
    radical surgery, but I hope nobody *intends* to just plain break the tree.
    To my knowledge, there are at least three folks doing simultaneous surgery
    on HEAD:

      - Martin refactoring the layout engine
      - Dom upgrading to GTK 2
      - you swapping in Pango

    So far, you've each been doing a great job of introducing your own changes
    in ways which hit CVS without affecting each other's ability to keep
    working. This is a very Good Thing.

    In particular, my understanding is that Martin is taking pains to ensure
    that at each step of his refactoring, he gets back to a known good state as
    quickly as possible.

    I shudder to think what would happen if you each felt free to break the tree
    for your own changes. Ouch. Having all three of you dead in the water or
    refusing to refresh your view against an unstable main repository sounds
    like a recipe for disaster -- or at least for some hellish merges

    I'm a realist, of course. There *will* be broken functionality and
    regressions here and there -- few refactorings are perfect up front -- but
    by staying disciplined, those can be rapidly fixed while the new code is
    fresh in the author's mind.

    >As far as the Pango work is concerned, this does not make it any
    >harder or easier, it just makes the sources clearer, particularly as
    >there is now going to be a new set of #ifdef WITH_PANGO
    >sections. In terms of progress, I have hardly started and doubt
    >there will be anything to show for it in the next month or two.

    Agreed. Getting the BIDI blocks out of the way will help make the PANGO
    demarcation a lot clearer.

    As I said in the prior message, after reviewing the code I'm totally sold
    now. Indeed, AFAICT, you're barely removing any non-BIDI code at all --
    mostly just all that ugly ifdef BIDI machinery.


    This archive was generated by hypermail 2.1.4 : Thu May 09 2002 - 19:30:07 EDT