Barak Pearlmutter -
> It would appear that if you use a losslevel of 100 (the recommended
> value, and the value used by -lossy) the problem disappears. The
> resulting file size increase is under 3%.
>From a user's point of view, I expect the documented range of
parameters to give me results ranging from comparatively big files
with lossless compression, to ugly output with compression too
aggressive. I don't expect dropped characters.
>From a programmer's point of view, it looks like an error. I.e. as I
understand the algorithm, the program makes one pass over the input
creating a dictionary of shapes, then outputs a series of pointers to
the dictionary. In this case it looks like it output an invalid
pointer or something.
> Unless you object (perhaps by suggesting a patch to the place in the
> documentation that led you to use 200 there) I would be inclined to
> mark this bug as closed.
I suggest a warning something like this:
--- cjb2.1-orig 2005-08-18 20:45:38.000000000 -0400
+++ cjb2.1 2005-08-18 20:51:26.000000000 -0400
@@ -91,6 +91,9 @@
Higher values generate smaller files
with more potential distortions.
Value zero corresponds to lossless encoding.
+Values above 100 may lead to character substitution errors, especially
+at resolutions above about 150 points per character (900 dpi for 12
+pt characters).
.TP
.B "-verbose"
Display informational messages while running.
- Jim Van Zandt
--
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
