Le Wed, May 22, 2002, à 04:35:32PM +0100, Tariq Rashid a écrit:
sorry - i've not got my emails for a few days... attached is a sample .dia file and the exported .eps ...
Tariq,
I can read both files with no problems. I'm using dia 0.90.RC2,
Ghostscript 7.04, Ghostview 3.5.8. The only little problem I see is that
half a letter is cropped off the left and right borders of the EPS. It seems
that it's a problem of the fudge factor/font bounding box, as usual (and we
have no clue, as usual. Lars? )
After running
ps2ps int_test_suite_network_net.eps toto.eps
the resulting toto.eps file (attached here) is indeed cropped down. The
original bounding box is 0 0 1032 1195, the final one is 0 0 595 842.
If we ignore the font fudge factor and assume your original EPS has the
correct bounding box (I sense two overlapping bugs here, one of which is
unknown and serious, the font stuff being finally minor but a bitch to
properly fix...), I really wonder what confuses ps2ps (I see it doesn't
output %%Magnification or %%Orientation fields; Hmmm. I tried removing them,
no dice).
We are issuing a pair of scale and translate commands at the beginning of
the EPS file; it could be a clue. Perhaps we aren't doing it the right,
Adobe-compliant way ? Or are we triggering a bug in ps2ps ?
Either way, I'm puzzled (and too tired to tackle this the gdb way this
evening).
Can you check whether you see an obvious relationship between the original
bounding box 0 0 1032 1195, the following Postscriptese
28.346000 -28.346000 scale
-4.815950 -46.101133 translate
and the resulting bounding box 0 0 595 842 ? perhaps this is the key...
(informally, the diagram's bounding box for dia is about 0 0 42.5 46 dia
centimetres (dcm))
(the ps2ps output issues a
0.1 0.1 scale
and then uses already translated units. I don't quite understand the
GS-generated garbage, but it seems that if the bounding box was fixed, the
diagram would be recovered. Now trying manually. Naah, doesn't work:
if you manually edit the ps2ps'd eps' bounding box, you get the original
size, but what's outside the smaller BB is clipped forever).
(we may want to output, as a debugging aid, the diagram's bounding box in
dcm, as a Postscript comment somewhere in the header !)
-- Cyrille
--
Grumpf.
Attachment:
toto.eps
Description: PostScript document