Re: [Devel] [slightly-OT] Non-integer bbox for T1 font?

[Tom Kacvinsky]

Does the customer's font file(s) (PFM + PFB, or suitcase + outline font) work
with ATM?  Or is the Type 1 font in question autogenerated?  For instance, on a
PC platform, the print drivers could be making (depending on how the driver is
configured) a Type 1 font out of a TT font (resulting in non-integer bboxes),
and these non-int bboxes are getting passed through to the PDF file when the PS
file is distilled.

If the Type 1 font is not autogenerated and works with ATM, we ought to add real
number parsing to the Type 1 driver's font bbox parsing code.

If the font is autogenerated, we have no way of really testing it with ATM, so
I am guessing in this case that Acrobat Reader (which uses CoolType, right?)
has more robust font bbox parsing.  Perhaps I could make a test suitcase and
outline file (using TransType from FontLab) and try that with ATM.  That would
help in testing.  I guess I need to see the PS file (was this made with one
of Digital Applications products?) and the original PDF file.

Tom

On Mon, 5 Mar 2001, Leonard Rosenthol wrote:

> Can anyone comment about the validity of a font where the bounding
> box is expressed in non-integer (aka real) values?  I found one in a
> PDF document that a customer of ours sent in and that causes both
> T1lib and FreeType to choke :(.
>
> Is this an invalid font (Acrobat seems to like it), or just unusual?
>
>
> Leonard
> P.S. File available on request
>