PICT OUTPUT?

["Stephen Ervin" <Stephen_Ervin@venus.gsd.harvard.edu>]

REGARDING                PICT OUTPUT?
I wonder if anyone else has experienced, or solved, the following problem?
I've used the pict-output code originally provided by bill st clair (back in
1.3.2 or therabouts, in a file called 'Picture-Files" or something
equivalent...) for some time, quite happily and successfully. BUT:

Now we have institutionally upgraded to using Adobe PhotoShop 2.5 everywhere, 
and Photoshop rejects all PICT files created by said code with a terse 'Problem
Parsing Pict' (its at least alliterative...) although older software (ie
PixelPaint Pro,  NIH Image, and others) still accept the PICT files with no
problem... (This all true whether BITMAP or OBJECT type graphics are included.)

A bit of digging and grubbing reveals the fact that 'extended version 2' PICT
files should have, after the first ten bytes of picture size,  a 2-byte
'Version OpCode', $0011, then followed by Version # $02FF,  and header opcode
$0C00 (p. 17-24,  IM VI, Table 17-4)...
The $02FF and subsequent bytes,  but not the $0011,  were being put into PICT
files --  I presume by the with-output-pict-file macro,  although that code is
sufficiently obscure to me that I dont really get it.  Nonetheless hacking that
code to insert the missing word after the pict size seemed to work,  and to
produce acceptable PICT files...

I say *Seemed* to work,  because I now am faced with that most annoying type of
bug:  intermittent and irreproducible.  *Sometimes*,  PICT files created this
way are still unreadable to fussy PhotoShop (often with a variant error
message:  "Not Enough Memory" -- I presume bogus,  since there's plenty.)  The
same save procedure executed twice in a row (on the same window, etc.) can
produce one good, and one bad, PICT file.... I even thought for a while that
writing directly to the DeskTop created bad files,  while writing into a folder
was ok,  since the tests showed that for a while,  but that was only
coincidence for a while... (and completely unsupportable by any hypothesis I
could come up with,  but am I wrong?)

At any rate,  does anyone have reasonably robust with-pict-ouput-file type code
that is know to work up to 'extended version 2' specs?
(OOU is little help,  since its PICT output macro is basically the same,  and
produces equally 'bad' files....) or any pointers (besides IM and defpascal???)
Thanx in advance...

PICTfully,

Stephen M Ervin
 Landscape Architecture/ Harvard U.

servin@gsd.harvard.edu