Reflecting on LUDEX

From: Tomas Frydrych (tomasfrydrych_at_yahoo.co.uk)
Date: Thu Apr 22 2004 - 02:52:47 EDT

  • Next message: Martin Sevior: "commit: Dragging left ruler controls looks nice."

    Hi guys,

    Going to LUDEX has been a worth-while experience. In particular, for
    it was an opportunity to meet face to face people who use AbiWord and
    reflect on their comments.

    There are two things that matter most to our userbase (1) stability,
    and (2) compatibility with Word. Stability in particular is a big
    issue. While there were number of people who came to see us and say
    'great software, use it and love it', there was a significant number
    of people who came and said 'I have tried AbiWord a while back and it
    kept crashing on me, so I gave up'. To my mind it is absolutely
    critical that when we release 2.2, it simply does not crash. I am not
    convinced that we can achieve that by GUADEC, although I believe we
    should have a solid beta by then; in any case, I would argue strongly
    that we should not release until all serious bugs in our bugzilla
    have been cleared out. I also think that we have to clean up all
    unhandled asserts from our code before we release it.

    The comments of various people who came along also made me think
    about where do we fit alongside OpenOffice. It has become clear to me
    that as a wordprocessor for a desktop/laptop system, our primary user
    base is, and likely will remain, the small home user; we are unlikely
    to make serious inroads into the corporate market in competition to
    OpenOffice in this area, or certainly not directly and on our own. I
    was also surprised how many people never heard of AbiWord. We need to
    be more creative in creating publicity for ourselves.

    There are, however, two areas in which not only we can compete with
    OpenOffice, but where we are without competition. The first of these
    is the ability to use AbiWord as a command line file conversion
    utility. This is something that we have not particularly bothered to
    emphasise in the past, yet the possibility of using AbiWord for
    example under Apache to serve various documents (e.g., Word) as
    html/xhtml is something that we should exploit to the full. While,
    naturally, we do not write AbiWord to be principally a command line
    tool, it is one way in which to draw attention to ourselves outside
    of our current userbase.

    The other, and much more important area, is that of embeded systems,
    and other systems with limited resources. Here we are not just an
    alternative to OpenOffice, but as far as embeded Linux is concerned
    we are the only thing available. Coupled with Word compatibility and
    our already excellent feature set, AbiWord is the obvious choice, and
    indeed, we have made couple of useful contacts at LUDEX in this area.
    On reflection, I think this should significantly inform the future
    development of AbiWord. For example, when adding new features,
    particularly features with significiant demands on either processing
    power or memory, we should design them in a manner that would make it
    feasible and easy to build AbiWord without them. It might be useful
    to further develop the plugin architecture to allow loading and
    unloading of plugins on demand. For example, instead of loading an
    imp/exp plugin at start up, we would load it only when the user is
    importing a document it supports and unload it once the document is
    parsed. This should not be too difficult to do. We might also want to
    review the existing codebase with a view of making it easier to
    reduce AbiWord further, if required. I am not suggesting we should
    simply turn AbiWord into a wordprocessor for embeded systems, but
    that we consider ways of making maximally friendly in that regard.

    We also had an interesting discussion with the author of Scribus, a
    DTP application for Linux. Scribus is looking really good, and there
    are some areas where we could directly benefit from it. For example,
    he has got a pretty decent support for shapes, and since he uses xml
    file format, we should examine if we could perhaps share object
    representation, etc.

    In any case these were two good days, and it was good to meet Hub in
    person (I am very grateful to Hub for coming; it would have been hard
    going on my own.) Time to get back to the bugzilla.

    Tomas



    This archive was generated by hypermail 2.1.4 : Thu Apr 22 2004 - 03:03:50 EDT