Re: commit: Layout horizontally with screen units.

From: Andrew Dunbar (hippietrail@yahoo.com)
Date: Fri Jan 24 2003 - 21:41:20 EST

  • Next message: Andrew Dunbar: "EncodingManager usage note"

     --- Martin Sevior <msevior@physics.unimelb.edu.au>
    wrote: > On Sat, 25 Jan 2003, Andrew Dunbar wrote:
    >
    > > --- Martin Sevior
    > <msevior@physics.unimelb.edu.au>
    > > wrote: > On Fri, 24 Jan 2003, Patrick Lam wrote:
    > > >
    > > > > On Sat, Jan 25, 2003 at 06:28:22AM +1100,
    > Martin
    > > > Sevior wrote:
    > > > >
    > > > > > I left that in in case we needed to revert.
    > > > >
    > > > > I think that bonsai/cvs are better at
    > capturing
    > > > this sort of information
    > > > > than leaving the code looking silly; ifdefs
    > make
    > > > the code harder to
    > > > > maintain, and it's not like anyone's going to
    > go
    > > > and fix up all the evil
    > > > > hacks left behind 6 months ago; these things
    > just
    > > > accumulate until no
    > > > > one remembers why they're there anymore.
    > > > >
    > > > > > In any case I think abiword in this mode is
    > much
    > > > more user-friendly
    > > > > > than when using layoutunits. Text does not
    > > > extend past the left margin
    > > > > > and there are no more disappearing runs in
    > full
    > > > justification.
    > > > >
    > > > > There's something I don't understand about
    > font
    > > > handling and layout units.
    > > > > I'm going to look into it this weekend and see
    > if
    > > > I can understand it
    > > > > more. Clearly text should not extend into the
    > > > margins, but neither should
    > > > > it wrap differently when printing.
    > > > >
    > > >
    > > > We're up against accmulated rounding errors in
    > > >
    > > > 1. Choosing fonts sizes at low resolution.
    > > > 2. Actual font widths on screen can be as small
    > as 4
    > > > pixels wide when the
    > > > properlay scaled (from a high resolution device)
    > > > would be say, 4.3 pixels
    > > > 3. Low resolution hinting which shifts scaled
    > fonts
    > > > in a undefined manner.
    > >
    > > Surely then the only ways to guarantee that screen
    > > text matches printed text is to scale screen fonts
    > > using subpixels but with hinting turned off so
    > that
    > > character widths don't vary at small sizes.
    > > But then we lose the nice look of small size fonts
    > -
    > > I'd only be willing to do this in Print Preview
    > mode.
    >
    > In the end we have to place glyphs in the screen at
    > integral values. The
    > only way to get close on screen to what will be
    > printed is the adjust EACH
    > glyph to be within 0.5 pixels of where it would be
    > on a high resolution
    > device. This is what Joaquin is arguing for and what
    > I agree with.

    The problem I was thinking of is with Arabic where all
    letters are supposed to join up exactly is we can
    never let a visible one pixel gap creep in here.
    Also, if two antialiased glyphs overlap by one pixel
    we need to ensure that the antialising is done after
    the overlapping or we will end up with artifacts.
    Perhaps these might be too tiny to notice - I couldn't
    say in advance.

    Andrew.

    > Cheers
    >
    > Martin
    >
    >

    =====
    http://linguaphile.sourceforge.net/cgi-bin/translator.pl http://www.abisource.com

    __________________________________________________
    Do You Yahoo!?
    Everything you'll ever need on one web page
    from News and Sport to Email and Music Charts
    http://uk.my.yahoo.com



    This archive was generated by hypermail 2.1.4 : Fri Jan 24 2003 - 21:44:19 EST