[Devel] Re: Distiller produced PFA problems

[Tom Kacvinsky]

Grrr....

I found the problem, finally.  There is bug in FreeType that reports there is no
CharStrings dictionary when there is one.  Adding a FontInfo dictionary causes
the false report to go away.  I'll see what I can come up with.  But it will
have to wait until tomorrow.  I have to help a friend with her computer, so away
I go...

Eric, I hope this isn't causing you too much stress, what with the 11th hour
show stopper you were referring to...

Tom

On Fri, 19 Jan 2001, Tom Kacvinsky wrote:

> Wow!  I am really stymied now.  I just finished looking at the font on one of
> the True64 4.0F boxes here at work.  No problems whatsoever.
>
> The tools I used to look at the "guts" of the font might have stripped DOS 
> EOLs
> (which may exist inside the eexec portion of a Type 1 font).  Or perhaps
> t1disasm converted Mac EOLs to UNIX EOLs.  I will investigate the influence of
> EOL on the line-by-line parser in FreeType.  This *might* be the hang up.
>
> One last thing to make sure we are on the same wavelength: did you add a dummy
> FontInfo dictionary to the font?  I don't recall you saying if you did this or
> not.  I had to do this in order to get ftview to grok the font.  Then again,
> adding a FontInfo dictionary doesn't account for failing because of a 
> supposedly
> missing CharStrings dictionary. :(
>
> Tom
>
> On Fri, 19 Jan 2001, Tom Kacvinsky wrote:
>
> > I'll try on a Compaq (DEC) Tru64 UNIX box tonight (using the native C 
> > compiler).
> > I'll let you know what I find out...  The byte sex issues aren't really an 
> > issue
> > for the Type 1 and CFF drivers.  Each specification states that 3 and 5 byte
> > encoded integers (1 escape byte, and either 2 or 4 data bytes) are to be in 
> > big
> > endian order, and the code currently follows those guidelines...
> >
> > Tom
> >
> > On Fri, 19 Jan 2001, Blinn, Eric W wrote:
> >
> > >
> > > Tom Kacvinsky wrote:
> > > >
> > > > I didn't mean to sound so snippy...
> > >
> > > No problem Tom.  I didn't think you were being snippy anyway.
> > > And I surely didn't mean to sound like I was snippily responding
> > > to your snippiness!  Okay, your turn.  ;-)
> > >
> > > So with just your patch applied to the current CVS snapshot, you were able
> > > to open the font with ftview just fine?  Hmm... that is interesting.
> > >
> > > Well, let me clarify one thing.  I don't have CVS here, so I am working 
> > > with
> > > the most recent freetype2-current and ft2demos-current snapshots I could
> > > get. It was timestamped 1/19/01 10:45 on the SourceForge FTP server.  I
> > > should think that has to be real close if not exactly the same source 
> > > tree.
> > > As further assurance, according to the ViewCVS web interface nothing in 
> > > the
> > > Type 1 driver has changed in at least two weeks.
> > >
> > > Now, as for the possibility of a platform dependent issue...  I'm 
> > > currently
> > > working on Windows NT using Visual C++ 5.0 and also DEC OSF/1 using the
> > > native DEC compiler.  On both platforms, ftview fails to open the PFA
> > > example I sent.  Both were built with the source snapshots I mentioned 
> > > above
> > > and your patch applied.  So that does not rule it out entirely, but it
> > > certainly suggests that this is not a platform specific issue (at least 
> > > not
> > > one single platform).
> > >
> > > Of course, another possibility is an endian-ness dependent issue.  Intel 
> > > and
> > > Alpha are little-endian.  SPARC is big-endian.  Perhaps a byte-order
> > > dependency?
> > >
> > > - Eric
> > >
> > > > -----Original Message-----
> > > > From: Tom Kacvinsky [mailto:address@hidden
> > > > Sent: Friday, January 19, 2001 8:03 PM
> > > > To: address@hidden
> > > > Subject: [Devel] Re: Distiller produced PFA problems
> > > >
> > > >
> > > > Hey...
> > > >
> > > > I didn't mean to sound so snippy...  What I was after is that
> > > > I did not
> > > > encounter the errors you are seeing.  Which is strange,
> > > > because we are using the
> > > > same code base, and the same font.  Which platform are you
> > > > working on?  Maybe
> > > > that is the issue?  I am using sun-sparc-solaris-2.7,
> > > > compiled with gcc 2.95.2.
> > > >
> > > > Tom
> > > >
> > > > On Fri, 19 Jan 2001, Blinn, Eric W wrote:
> > > >
> > > > >
> > > > > Tom Kacvinsky wrote:
> > > > > >
> > > > > > The font you sent me has a CharStrings array!  Did you
> > > > > > send me the wrong font?  I am working with the latest CVS
> > > > > > version, so if you are using an earlier version of the code,
> > > > > > that might be the cause of the problem...
> > > > >
> > > > > Hey Tom,
> > > > >
> > > > > No.  You have the right font and I patched the right
> > > > version.  The problem
> > > > > is related to the point I was making in the e-mail I sent
> > > > with the font
> > > > > attached ...
> > > > >
> > > > > The subrs array is not all that is missing here.  The font
> > > > does not have a
> > > > > charstrings array either, or at least one is not loaded by
> > > > the time we get
> > > > > to T1_Open_Face().  This doesn't satisfy T1_Open_Face().
> > > > Immediately after
> > > > > the check for a subrs array a similar check is made for a
> > > > charstrings array.
> > > > > It is not present and so error still gets set to
> > > > T1_Err_Invalid_File_Format
> > > > > and eventually propagates up to a
> > > > FT_Err_Invalid_File_Format coming out of
> > > > > FT_New_Face().
> > > > >
> > > > > - Eric
> > > > >
> > > >
> > > >
> > > > _______________________________________________
> > > > Devel mailing list
> > > > address@hidden
> > > > http://www.freetype.org/mailman/listinfo/devel
> > > >
> > >
> >
> >
> > _______________________________________________
> > Devel mailing list
> > address@hidden
> > http://www.freetype.org/mailman/listinfo/devel
> >
>
>
> _______________________________________________
> Devel mailing list
> address@hidden
> http://www.freetype.org/mailman/listinfo/devel
>