gdi32: GetGlyphIndices doesn't substitute glyph incorrect

[Aric Stewart]

Hi there,

 I know I am a year late to the party here but I am digging deep into tategaki for a few Japanese programs that are having serious rendering issues and I believe that this patch is incorrect.

I am attaching the test patch that proves it to me. Kusanagi-san, we should try to make sure that this does not break what it was that you where trying to fix with your original patch.

-aric

[Kusanagi Kouichi]

On 2013-05-10 09:23:17 -0500, Aric Stewart wrote:
> Hi there,
>
> I know I am a year late to the party here but I am digging deep into tategaki for a few Japanese programs that are having serious rendering issues and I believe that this patch is incorrect.

Can you explain the problem? How do I reproduce it? I know that
tategaki has a minor issue, but it's not serious. Does reverting
commit fecb1d8b6f81c2bfff7116e80380b1d4f7b7befc ("gdi32:
GetGlyphIndices doesn't substitute glyph.") help?
> I am attaching the test patch that proves it to me. Kusanagi-san, we should try to make sure that this does not break what it was that you where trying to fix with your original patch.

Your test passed here.
> -aric

[Aric Stewart]

Hello Kusanagi-san,

On 5/15/13 6:03 AM, Kusanagi Kouichi wrote:
> On 2013-05-10 09:23:17 -0500, Aric Stewart wrote:
> > Hi there,
> >
> > I know I am a year late to the party here but I am digging deep into tategaki for a few Japanese programs that are having serious rendering issues and I believe that this patch is incorrect.
>
> Can you explain the problem? How do I reproduce it? I know that
> tategaki has a minor issue, but it's not serious. Does reverting
> commit fecb1d8b6f81c2bfff7116e80380b1d4f7b7befc ("gdi32:
> GetGlyphIndices doesn't substitute glyph.") help?

Just reverting the changes to GetGlyphIndices causes your original tests to fail. so I will have to investigate why and try to correct that. Did you have a specific test case that required this change?
> > I am attaching the test patch that proves it to me. Kusanagi-san, we should try to make sure that this does not break what it was that you where trying to fix with your original patch.
>
> Your test passed here.
> > -aric

I also thought that tategaki was mostly working, but I started to investigate and have found quite a number of serious issues.

I have attached my test program as well as png screen shots from win7 and wine to illustrate the issues I am trying to solve. Most interesting to me are the rotated glyph origin problem as well as the fact that @fonts rotate many of the characters by 90 degrees.

I have some work in progress that is fixing the origin, the incorrect GSUB substitution when using glyph indices and also the rotation.

I am finding the logic on how windows decides which glyphs get the 90 degree rotation and which do not very hard to figure out. Do you have any understanding on how that works?

thanks!
-aric

vertical.ttf

[Aric Stewart]

Hello Kusanagi-san,

I am trying to do more testing with tategaki and i am finding that your original tests are incorrect because of the vertical.ttf font you have build.

Windows appears to be very very picky about when it will to a vert or vrt2 replacement. I have run the test with MS UI Gothic, and i get behavior where "@MS UI Gothic" of U+2025 is a different glyph than for just "MS UI Gothic" contrary to your tests.

I have spent a day or so trying to get fontforge to build a ttf font that windows will do vrt2 or vert glyph replacements on and I cannot get it to work. Nothing I build seems to meet the requirements.

I am thinking of abandoning the vertical.sfd and ttf and instead just running the test with MS UI Gothic or something else that is a font that Microsoft properly does vert replacements on.

Do you have any insight or help here that we could do to make these tests function properly?

thanks!
-aric

vertical.ttf

[Kusanagi Kouichi]

On 2013-05-16 12:01:21 -0500, Aric Stewart wrote:
> Hello Kusanagi-san,
>
> I am trying to do more testing with tategaki and i am finding that your original tests are incorrect because of the vertical.ttf font you have build.
>
> Windows appears to be very very picky about when it will to a vert or vrt2 replacement. I have run the test with MS UI Gothic, and i get behavior where "@MS UI Gothic" of U+2025 is a different glyph than for just "MS UI Gothic" contrary to your tests.
>
> I have spent a day or so trying to get fontforge to build a ttf font that windows will do vrt2 or vert glyph replacements on and I cannot get it to work. Nothing I build seems to meet the requirements.
>
> I am thinking of abandoning the vertical.sfd and ttf and instead just running the test with MS UI Gothic or something else that is a font that Microsoft properly does vert replacements on.
>
> Do you have any insight or help here that we could do to make these tests function properly?
>
> thanks!
> -aric

Sorry, I cannot help right now. I'll investigate this.