Win32 Stable char widths still wrong!

From: Ryan Pavlik (
Date: Thu Jan 22 2004 - 22:12:12 EST

  • Next message: "commit: Fix 6331 and some more stuff for TOC."

    Oh, what changing the zoom level then typing some papers can do to
    reveal bugs...

    On uwog's request I am sending this along to the list. For those of you
    (like me) on dialup connections, I have not sent along all the
    screenshots. The total is 223 KB, so this should take less than 3
    minutes on dialup. Those of you who want them should email me (not
    cc'ing the list, for sanity) and I'll fling them along. Eventually I
    will find the bug to put these with, but uwog had a sense of urgency,
    that I share whenever my zoom level is != 100x where x is a positive

    And there I go again, writing a lot, when I promised just to take shots
    and let them speak for themselves. So here are the comments for the shots.

    Win201100pct: Stock install of 2.0.1 behavior at 100 pct. Shows no bugs
    of this sort, because they all disappear when zoom is an exact multiple
    of 100.

    Win201150pct: Stock install of 2.0.1 behavior at 150 pct. Shows
    discrepancies in width. Insertion pointer is not displayed, but rest
    assured it was misplaced (just past the 6, as it was supposed to be
    right at the end of the line.)

    Winstable0122100pct: Tonight's nightly build, roll-your-own by me,
    mingw, well cleaned. Doesn't show bug, as it's 100 pct zoom.

    Winstable0122150pct: This one shows it. Big time. Note location of
    speeling error underline, displayed insertion pointer (it was logically
    at the end of the line, as I just typed that right in. Note also how
    the text floods past the margin. There are redrawing issues when it
    does that (had to abbreviate "something" to get it to go to the end of
    the first line and do that).

    Winstable0122150pctinsert: Here I typed a bunch of stuff ("This is
    existing text, oh look at the beauty of the fact that is exists. Now
    watch as I edit."), then clicked at what appears to be the end of "fact"
    and typed my little inserted sentence. Not where it was expected to be.
      Note also where the spelling squiggle for "thatI" (a product of the
    insertion) is.

    This is a set of three.
    stabpic = "Winstable0122150pct" // I am bored of typing that so much.

    stabpicpointerendofline: This is the text stating what I'm gonna do.
    Note insertion pointer. I just typed this straight in.

    stabpicmidselection: I begin to do what I stated I was going to do in
    the above shot. Note how weird what was selected was, and what happened
    to its positioning.

    stabpiccompletedselection: I finish dragging to what used to be the
    beginning of the sentence, onscreen. Note what was actually selected.

    A few comments on my environment:
    WinXP pro. Freaky color scheme (matches a desktop I made), but sizes
    are all left default (can't stand pixelated icons).
    Fonts are normal to the best of my knowledge. When bug testing I tend
    to use Times New Roman if it works to show the bug, and it does, so I did.
    I left out a bunch of 100pct shots since they all look like they're
    supposed to.

    And a few words from uwog, courtesy of #AbiWord (where all the cool kids
    hang out...):
    <uwog> is that in STABLE or HEAD?

    <RyanPavlik> It looks like it's in both, but worse in head
    * RyanPavlik was so desperate last week he used 1.0.5
    <uwog> that sucks massive
    <uwog> please report anything you find to the ML
    <RyanPavlik> alright
    <uwog> so jordi or someone else can look into it
    <uwog> should be fixes sooner than later
    <RyanPavlik> All I know is that it works just fine in 100% and 200%,
    just like all those other weirdo variable-width font issues
    <RyanPavlik> It increases in severity as the line goes from left to
    right, so I'm figuring it's something that's being looped or called
    <uwog> most likely, there is still a bug in the Win32 Char measuring
    width routine

    <uwog> dom fixed it for STABLE
    <uwog> so I'm gonna kill someone if it broke in STABLE again
    * RyanPavlik ducks
    <RyanPavlik> I think it is

    <RyanPavlik> Do you recall when Dom patched it?
    <uwog> _just_ before 2.0.1
    <uwog> 2.0.0 did not have the patch
    <uwog> 2.0.1 did
    <RyanPavlik> let's see, 2.0.1 was a long time ago.
    <uwog> and afterwards no-one was allowed to break it :)

    <RyanPavlik> 2.0.1: cursor leads, ahead of text
    <uwog> so it's in 2.0.1 as well?
    <RyanPavlik> yes
    <RyanPavlik> but it's not as bad

    <uwog> RyanPavlik: could you write some 'status'mail, reporting the
    results with different versions/which fonts you use/which font sizes &&
    zoom settings ?
    <uwog> that would be nice
    <RyanPavlik> alrighty
    <RyanPavlik> to the devlist?
    <uwog> please, yes
    <uwog> and when it got worse
    <RyanPavlik> alrighty
    * uwog wonders why the win32 developers never notices/fixed/reported this
    <RyanPavlik> stable is looking really weird to me right now
    <uwog> s/notices/noticed

    And that's plenty long enough. I'll try to get it up on a bug tonight.
      I think I actually already have a variable-width font one that i've
    commented on, so I'll probably stick it all there.

    Sorry to bother you guys, but it is quite a bit of a usability problem,
    and telling win users to use 100 pct zoom probably won't go over too
    well with those with large, hi res monitors.


    This archive was generated by hypermail 2.1.4 : Thu Jan 22 2004 - 22:13:25 EST