> From: Andrew Smith
> >i don't understand why you are being so hostile
> Sorry, that wasn't my intention. I was trying to
analyse
> whether there is a problem in the absence of an
example,
FWIW, I didn't sense any hostility, either.
> however:
> >i even made an image that will illustrate the
problem. it's
> >toucan.png from the png book with an extra, empty
IDAT added in between
> >the last IDAT and IEND.
> >
> >http://l
ittlesvr.ca/misc/andorbug-toucan.png
> Which disproves my assertion... That's a valid PNG
yet at least
Indeed it is. In fact, any number of zero-length IDATs can
be in
there--start, middle, or end--as long as all of the IDATs
are contiguous.
This is clearly a libpng bug, albeit not necessarily one
with earth-
shaking ramifications.
> xv 3.10a displays it, but it encounters the error
(which it
> ignores, I guess that's ok for an image editor). xv
is using
> libpng-1.2.8
I think it's doing that more in its image-viewer role. XV
attempts
to display whatever's there, including corrupted data
(which happens
in GIFs fairly regularly and in JPEGs with some card readers
:-( ).
Most users expect/prefer this behavior, I think.
> I guess a minimal fix would be to remove the
PNG_AFTER_IDAT
> part of the test, but that would allow zero length IDAT
after
> a non-IDAT chunk.
Yeah, it really needs to track the contiguity of IDATs
explicitly.
(pngcheck does so and reports no errors, btw.)
Greg
-------------------------------------------------------
All the advantages of Linux Managed Hosting--Without the
Cost and Risk!
Fully trained technicians. The highest number of Red Hat
certifications in
the hosting industry. Fanatical Support. Click to learn more
http://sel.as-us.falkag.net/
sel?cmd=lnk&kid=107521&bid=248729&dat=121642
_______________________________________________
png-mng-implement mailing list
png-mng-implement lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/png-m
ng-implement
|
