[pdftex] Patch for pdftex for subsetted font printing problem

[Tom Kacvinsky]

Hi Mark,

I meant to write you about this whole morass to make sure that dvipdfm
didn't "suffer" from the same problem.

Despite the published specification, I have been told that rand is a
supported Type 1 operator. :(

I didn't say it was impossible to take into account such things as div, nor
was I trying to give the impression that it was impossible to get pdftex to
renumber charstrings.  I was trying to get at the fact it would take more
time and care than I am able to expend right now, esp. with the possibilty
of calculated subr numbers.  I am sure you know the feeling.

I guess you are right.  Perhaps the best thing to do is to not subset subrs.

No, I don't think seac can be used random arguments, unless one is really
crafty and gets operans right (remeber tht there are some strict rules on
what the operands to seac mean, like certains have to match, etc...). I do
not think this likely.

No, Adobe is not reccommending that we not subset subrs.  They were
pointing out some pitfalls of subsetting subrs, and where in the Type 1
spec it says to avoid gaps in the Subrs array.  That tidbit is in the
section on ATM compatibilty, not of much use for PS, which is more general
than ATM.

Tom

On Mon, 18 Jun 2001, Mark A. Wicks wrote:

>
>
> > On that note, I have to ask: how does DVIPSONE get around this?  Does it
> > fully interpret the charstirng commands?  Did you know that rand is a
> > supported Type 1 charstring command?  How do you handle that, especially
>
>     That's interesting.  I didn't know "rand" was a supported Type 1
> charstring command. It's not documented in the "Adobe Type 1 Font Format."
> If the "Adobe Type 1 Font Format" is a complete list of charstring
> commands, the only commands that seem to be a problem are div,
> callothersubr, callsubr, and pop.  If that's true, it doesn't sound
> impossible to partially interpret "div" and a few other operators.
>
>     In any case, if "rand" is supported, the only safe thing would be not
> to subset subrs. Is that what you are suggesting?
>
> > I am taking my cue from the email we (Thanh, you, and I) received this
> > weekend from Adobe...
>
>    Is Adobe recommending not to subset subrs?  If "rand" is available in
> charstrings, then could "seac" also get called with random arguments?
> That would argue against subsetting charstrings too.
>