Re: [ft] retrieving kerning infos misbehaves

2007-03-08 Thread Jürgen Herz
David Turner wrote:

> not coïncidence, I just had a quick look at the code after your
> initial e-mail and saw the typo. Didn't you see my e-mail on the
> mailing list about it ?

No sorry, also can't find it neither on gmane nor the lists web archive.
But anyways, thanks for the fix.

>> BTW2: In order to get an debugable build I wanted to use the
>> developer target but didn't manage it. I didn't find explicit
>> instructions how to do that. Finally I moved gcc-dev.mk and
>> unix-dev.mk to gcc.mk and unix.mk and it worked. But how is it done
>> the smart way?
>>
>
> simply type "make setup devel" :-)

Maybe that's to easy ... Thanks.

Jürgen


___
Freetype mailing list
[email protected]
http://lists.nongnu.org/mailman/listinfo/freetype


Re: [ft] retrieving kerning infos misbehaves

2007-03-07 Thread David Turner
Hi

On Thu, 08 Mar 2007 00:08:43 +0100, "Jürgen Herz" <[EMAIL PROTECTED]> said:
> Thanks for fixing this bug (whoever that did) two days ago (rev 1.114 of
> ttobjs.c). Was this coincidence through my first report?
>
not coïncidence, I just had a quick look at the code after your initial
e-mail and saw the typo. Didn't you see my e-mail on the mailing list about it ?
 
> BTW: size->cvt_ready is queried twice but never set to anything else
> than 0. Is it really necessary to always call tt_size_ready_bytecode()
> in tt_loader_init().
> 
Will check that tomorrow

> BTW2: In order to get an debugable build I wanted to use the developer
> target but didn't manage it. I didn't find explicit instructions how to
> do that. Finally I moved gcc-dev.mk and unix-dev.mk to gcc.mk and
> unix.mk and it worked. But how is it done the smart way?
>

simply type "make setup devel" :-)
 
Hope this helps,

- David


> Thanks,
> Jürgen
> 
> 
> ___
> Freetype mailing list
> [email protected]
> http://lists.nongnu.org/mailman/listinfo/freetype


___
Freetype mailing list
[email protected]
http://lists.nongnu.org/mailman/listinfo/freetype


Re: [ft] retrieving kerning infos misbehaves

2007-03-07 Thread Werner LEMBERG

> Thanks for fixing this bug (whoever that did) two days ago (rev
> 1.114 of ttobjs.c). Was this coincidence through my first report?

It was coincidence -- the other report was easier to comprehend.

> BTW: size->cvt_ready is queried twice but never set to anything else
> than 0.

This is a bug.  David?

> Is it really necessary to always call tt_size_ready_bytecode() in
> tt_loader_init().

The CVT code must be rerun if the ppem is changing.

> BTW2: In order to get an debugable build I wanted to use the
> developer target but didn't manage it. I didn't find explicit
> instructions how to do that. Finally I moved gcc-dev.mk and
> unix-dev.mk to gcc.mk and unix.mk and it worked. But how is it done
> the smart way?

  [make distclean]
  make devel
  make


 Werner


___
Freetype mailing list
[email protected]
http://lists.nongnu.org/mailman/listinfo/freetype


Re: [ft] retrieving kerning infos misbehaves

2007-03-07 Thread Jürgen Herz
Werner LEMBERG wrote:
>> > It's weird, the first call to FT_Get_Kerning() always fills in the
>> > right value for a kerning pair, regardless of any flag. [...]
> 
> You would help us a lot if you can provide a small, complete,
> compilable C code example which demonstrates the problem.  And please
> use the latest CVS for testing.

I found the buggy line myself today by running the lib in the
debugger--but someone else was faster.

Thanks for fixing this bug (whoever that did) two days ago (rev 1.114 of
ttobjs.c). Was this coincidence through my first report?

BTW: size->cvt_ready is queried twice but never set to anything else
than 0. Is it really necessary to always call tt_size_ready_bytecode()
in tt_loader_init().

BTW2: In order to get an debugable build I wanted to use the developer
target but didn't manage it. I didn't find explicit instructions how to
do that. Finally I moved gcc-dev.mk and unix-dev.mk to gcc.mk and
unix.mk and it worked. But how is it done the smart way?

Thanks,
Jürgen


___
Freetype mailing list
[email protected]
http://lists.nongnu.org/mailman/listinfo/freetype


Re: [ft] retrieving kerning infos misbehaves

2007-03-07 Thread Werner LEMBERG

> unfortunately I got no replies on my previous posts.  Seeing
> Johannes' post gets some replies and even a bug was discovered I've
> new hopes regarding my issue and try it again.

Be patient.  Basically, it's only David and me who are working on
FreeType, and we have a lot of other things to do.

> > It's weird, the first call to FT_Get_Kerning() always fills in the
> > right value for a kerning pair, regardless of any flag. [...]

You would help us a lot if you can provide a small, complete,
compilable C code example which demonstrates the problem.  And please
use the latest CVS for testing.


Werner


___
Freetype mailing list
[email protected]
http://lists.nongnu.org/mailman/listinfo/freetype


Re: [ft] retrieving kerning infos misbehaves

2007-03-06 Thread Jürgen Herz
Hi,

unfortunately I got no replies on my previous posts. Seeing Johannes'
post gets some replies and even a bug was discovered I've new hopes
regarding my issue and try it again.

Jürgen Herz wrote:
> It's weird, the first call to FT_Get_Kerning() always fills in the right
> value for a kerning pair, regardless of any flag.
> 
> After a call to FT_Load_Glyph() later calls to FT_Get_Kerning() only
> give the right value if it has FT_KERNING_UNFITTED or
> FT_KERNING_UNSCALED set. Or when called with FT_KERNING_DEFAULT if
> FT_Load_Glyph() has FT_LOAD_NO_HINTING or FT_LOAD_NO_SCALE (which
> includes FT_LOAD_NO_HINTING) set.
> 
> But is FT_KERNING_DEFAULT set in FT_Get_Kerning() and FT_Load_Glyph()
> has either FT_LOAD_DEFAULT, FT_LOAD_RENDER or FT_LOAD_NO_BITMAP set,
> kerning infos are always 0.

I've to add that this only applies to TrueType fonts, Type1 fonts seem
not to have this issue.

Regards,
Jürgen


___
Freetype mailing list
[email protected]
http://lists.nongnu.org/mailman/listinfo/freetype