Re: GDI32: Fix logical error of italic,bold determination

[Huw D M Davies]

On Mon, Mar 13, 2006 at 04:10:04AM +0900, Byeong-Sik Jeon wrote:
> ChangeLog:
> Fix the logical error of the italic, bold determination.
>
> Index: dlls/gdi/freetype.c
> ===================================================================
> RCS file: /home/wine/wine/dlls/gdi/freetype.c,v
> retrieving revision 1.114
> diff -u -u -r1.114 freetype.c
> --- dlls/gdi/freetype.c	9 Mar 2006 12:30:52 -0000	1.114
> +++ dlls/gdi/freetype.c	12 Mar 2006 10:01:22 -0000
> +    for(i = 0; !face && i < 4; ++i) {
>          best = NULL;
>          LIST_FOR_EACH(face_elem_ptr, &family->faces) {
>              face = LIST_ENTRY(face_elem_ptr, Face, entry);
> -            if((csi.fs.fsCsb[0] & (face->fs.fsCsb[0] | face->fs_links.fsCsb[0])) || !csi.fs.fsCsb[0]) {
> +            if((!(face->Italic ^ it) ^ ((i >> 1) & 1)) && (!(face->Bold ^ bd) ^ (i & 1)) &&
> +               ((csi.fs.fsCsb[0] & (face->fs.fsCsb[0] | face->fs_links.fsCsb[0])) || !csi.fs.fsCsb[0])) {

It doesn't seem right to compare these with 'i', the order of the faces is random.

The right way to do this is to rewrite all this font matching code so
that we loop through the families and faces once and score each one.
At the end we pick the one with the lowest score.

Huw.
-- 
Huw Davies
huw(a)codeweavers.com

GDI32: Fix logical error of italic,bold determination

[Byeong-Sik Jeon]

2006-03-15 (수), 14:50 +0000, Huw D M Davies 쓰시길:
> On Mon, Mar 13, 2006 at 04:10:04AM +0900, Byeong-Sik Jeon wrote:
> > ChangeLog:
> > Fix the logical error of the italic, bold determination.
> >
> > Index: dlls/gdi/freetype.c
> > ===================================================================
> > RCS file: /home/wine/wine/dlls/gdi/freetype.c,v
> > retrieving revision 1.114
> > diff -u -u -r1.114 freetype.c
> > --- dlls/gdi/freetype.c	9 Mar 2006 12:30:52 -0000	1.114
> > +++ dlls/gdi/freetype.c	12 Mar 2006 10:01:22 -0000
> > +    for(i = 0; !face && i < 4; ++i) {
> >          best = NULL;
> >          LIST_FOR_EACH(face_elem_ptr, &family->faces) {
> >              face = LIST_ENTRY(face_elem_ptr, Face, entry);
> > -            if((csi.fs.fsCsb[0] & (face->fs.fsCsb[0] | face->fs_links.fsCsb[0])) || !csi.fs.fsCsb[0]) {
> > +            if((!(face->Italic ^ it) ^ ((i >> 1) & 1)) && (!(face->Bold ^ bd) ^ (i & 1)) &&
> > +               ((csi.fs.fsCsb[0] & (face->fs.fsCsb[0] | face->fs_links.fsCsb[0])) || !csi.fs.fsCsb[0])) {
>
> It doesn't seem right to compare these with 'i', the order of the faces is random.
>
> The right way to do this is to rewrite all this font matching code so
> that we loop through the families and faces once and score each one.
> At the end we pick the one with the lowest score.
>
> Huw.

Hi. I'm late...

A   B    A xor B
0   0       0
0   1       1
1   0       1
1   1       0
===>   0 ^ A == A ^ 0 == A,   1 ^ A == A ^ 1 == !A

This code is just a compression of my old patch [3/5].
but, this dirty code did not satisfy me too.


I will rework this and others(FT_Bitmap_Convert, FT_Sfnt_Name).
Thanks your advice. :)