🚀 go-pugleaf

Re
<9204111911.AA11232@raisin-nut>
<THOMAS.92Apr11192817@sorley.cogsci.ed.ac.uk>:

> For what they are worth, here are some observations I made with
> respect to the problem, using the real mode MSDOS executable of
> Ghostscript2.4 (gs.exe):
> 
> (a) I printed a document to disk using the new Windows 3.1 PS driver.
> The driver downloads an error handler into the print-file which
> will print out stuff when things go haywire, e.g. when the interpreter
> can't successfully load the font file. Now, the interpreter has to
> print these messages and it does so (hold your breath) using the very
> same font it choked on.(??)
> 
> (b) While the interpreter rejects some (but not all) dubious converted
> binary font files, it will also barf at some (but not all) Adobe fonts
> (see first quote).
> 
> (c) The protected mode version of gs24 (gs386.exe) will load all of my
> fonts, _including_ those rejected by gs.exe with no problems at all.
> (I thought both versions used the same code, or do they?)  Alas,
> gs386.exe and Windows will not coexist (Unix people stop sniggering
> :-} ).

I suspect the following:

	- The non-386 version does not use extended or expanded
memory, and it typically runs out of memory at about 3 or 4 fonts.
If the error name is VMerror, that's what's happening.

	- There is a bug in the non-386 version which causes it to
get confused and mis-parse .PFB files if the eexec (encrypted
outline) portion of the file is longer than 32K bytes and shorter
than 64K bytes, or longer than 96K and shorter than 128K.  I suspect
a number of Adobe fonts have this property.  This bug will be fixed
in 2.4.1, scheduled for release on April 16.

L. Peter Deutsch :: Aladdin Enterprises :: P.O. box 60264, Palo Alto, CA 94306
ghost@aladdin.com, ...decwrl!aladdin!ghost ; 415-322-0103 8-9 AM M-F/msg
	    "Implementation is the sincerest form of flattery."