From: Donal K. Fellows on
On 29 Apr, 00:13, Jeff Godfrey <jeff_godf...(a)pobox.com> wrote:
> Great suggestion - thanks.  Yes, rotating the string to 0.01 does
> exhibit the originally reported problem.  That is, the text grows in
> exactly the same way (and in the same amount) as the 90 degree rotation.

OK, so it's getting the font size wrong when applying the rotation.
*bad*word* :-(

> The originally reported clipping issues are evident in this case
> also.

That's expected given what you reported above. Tk calculates the
metrics unrotated and then rotates them (it's easier than calculating
metrics directly with rotated fonts).

Donal.
From: Alexandre Ferrieux on
On Apr 29, 9:48 am, "Donal K. Fellows"
<donal.k.fell...(a)manchester.ac.uk> wrote:
>
> That's expected given what you reported above. Tk calculates the
> metrics unrotated and then rotates them (it's easier than calculating
> metrics directly with rotated fonts).

Then could it be that the rotated font has a different point-size-to-
actual-size ratio than the original ? At 45 degrees, the bounding box
of a given square character is sqrt(2) times larger than the original,
so you need a bitmap that much larger...

-Alex
From: Donal K. Fellows on
On 29 Apr, 09:48, Alexandre Ferrieux <alexandre.ferri...(a)gmail.com>
wrote:
> Then could it be that the rotated font has a different point-size-to-
> actual-size ratio than the original ?

If that's the case, then either we're asking for the wrong thing
somewhere, or there's a bug in Windows's font system, or there's a bug
in the Consolas font itself. I can't really pursue this though; I've
no Win7 system and back when I wrote the code it appeared to work when
I was spinning text in WinXP. (I have no idea whether I tested with
the Consolas font.)

Donal.
From: Alexandre Ferrieux on
On Apr 29, 12:47 pm, "Donal K. Fellows"
<donal.k.fell...(a)manchester.ac.uk> wrote:
> On 29 Apr, 09:48, Alexandre Ferrieux <alexandre.ferri...(a)gmail.com>
> wrote:
>
> > Then could it be that the rotated font has a different point-size-to-
> > actual-size ratio than the original ?
>
> If that's the case, then either we're asking for the wrong thing
> somewhere, or there's a bug in Windows's font system, or there's a bug
> in the Consolas font itself.

Yes, but waiting for Windows to fix itself is not in our roadmap,
right ?

On the workaround side, can you explain why "it's easier than
calculating metrics directly with rotated fonts" ? Is it hard to call
GetTextMetrics() on the rotated font ?

-Alex
From: Jeff Godfrey on
On 4/29/2010 5:47 AM, Donal K. Fellows wrote:
> On 29 Apr, 09:48, Alexandre Ferrieux<alexandre.ferri...(a)gmail.com>
> wrote:
>> Then could it be that the rotated font has a different point-size-to-
>> actual-size ratio than the original ?
>
> If that's the case, then either we're asking for the wrong thing
> somewhere, or there's a bug in Windows's font system, or there's a bug
> in the Consolas font itself. I can't really pursue this though; I've
> no Win7 system and back when I wrote the code it appeared to work when
> I was spinning text in WinXP. (I have no idea whether I tested with
> the Consolas font.)

I haven't tried anything but the Consolas font. I'll check out some
other fonts today and report back. Also, I haven't tried this under XP
as I don't have an 8.6 installation on an XP box. I'll remedy that
today and give it a try.

Jeff