Re: [Devel] Adding a new function, question about face's allocedmemory

[Tom Kacvinsky]

Hi Antoine,

You have convinced me that I should make the char code parameter be a FT_ULong.
The functions I will write for the Type 1/CFF drivers will return a NULL string
if the char code parameter is outside the range 0 -- 255.  Does this sound like
an acceptable compromise?

I will later work on the cmap synthesizer for the CFF driver.

Regards,

Tom

On Tue, 6 Nov 2001, Antoine Leca wrote:

> Hi Tom,
>
> Tom Kacvinsky wrote:
> >
> > Hi Antoine,
> >
> > (where were you at in Amsterdam?)
>
> To make a long and painful story short, I went, but I was unable to read
> my mail the day before, so I did miss the meeting point. And at 12 o'clock,
> there were nobody at the station... I turned back at 3 pm, because I were
> a bit ill and sick; I neglect to enter a cyberplace to check my mail, which
> it later appears to be an error :-(
>
>
> > > > The idea is for this function to be used with non-CID keyed CFF fonts
> > > > and Type 1 fonts (hence, the char code is always in the range 0 -- 255,
> > > > which is why I chose FT_Byte for the char_code parameter).
> > >
> > > Hmmm, this is the kind of "optimizations" I do not find useful.
> > >
> > > I see the same function as useful for type 4.0 'post' tables in 'sfnt'
> > > (TrueType) fonts, where the names of the characters, highly repetitive
> > > (like a0101, a0102, a0103, etc.) are encoded as 16-bit quantities;
> > > however, there may be as much as 65,536 characters in such a font, so
> > > your function won't do the job. :-(
> > >
> >
> > I guess you did not read that this function is for (pure) non-CID CFF and
> > Type 1 fonts, which can only encode 256 glyphs at a time.
>
> I did (and I know the fact), but I miss what is the planned status of your
> function. If it is to be specific to T1/non-CID, I then miss how it
> integrates in the Freetype 2 framework (but I must confess I did not toy
> with Freetype 2).
>
> At the contrary, if this is intended to be exposed at the "normal", visible
> side of the library, i.e. the API, then I believe we should _permit_ to
> re-use this entry point for similar uses, whatever they may be; and this
> means we should never restrict parameter ranges, particularly when it does
> not cost anything.
>
>
> > Unless you were proposing that I write this function to additionally get
> > glyph names for a given char code when working with TT fonts.
>
> No, I do not propose that. I only propose that it should be made possible
> in the future, if such a need do come.
> I defer on David commenting if my position does make sence, or not.
>
>
> > In which case, I have to take into account cmaps.  Which is what
> > I wanted to avoid in the first place...
>
> :-)  Yep, I did caught that very well...
>
>
> Antoine
>
> _______________________________________________
> Devel mailing list
> address@hidden
> http://www.freetype.org/mailman/listinfo/devel
>

-- 
all alone the life you lead -- a silent diner where you feed
bow your head and pretend to read -- this one is for you

Brian and Robert (from The Story of the Ghost, by Phish)