-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1
I just noticed a few things while testing some content in 1.2b9 and
the parser that comes with it.
1.) PIL doesn't natively support 8bpp in python-imaging version
1.1.3 from Debian Unstable. This was tested using python-2.1.3
and python-2.2.1.
2.) _ALL_ the image convertors seem to barf on a pretty large
majority of images, resulting in the same type of errors:
Error: Runtime error parsing document
http://www.cnn.com/.../tz.bush.sat.jpg:
Error while converting image.
Some images work, some do not. Increasing verbosity only seems
to report more information for netpbm, but not PIL or
imagemagick.
3.) Setting the distiller to 4bpp, then fetching the same site to
the same depth (NewsBlaster in my case) results in the familiar
[img][img] tags. There are images inside this document, because
I've seen them converted by the distiller. Changing the viewer
settings does not seem to expose the images, they're
"blackholed" inside the pdb. This has been discussed before in
the context of "removing" those [img] tags, but what about
replacing them with the images which clearly exist in the .pdb
file itself? Why aren't the images visible? What are the
conditions under which [img] is shown instead of an image?
4.) Changing the image parser to imagemagick still tries to find and
use ppmtoTbmp, but of course that doesn't exist anymore.. and it
won't do 8bpp or 16bpp anyway in the "standard" packaged
version without patching.
5.) While using netpbm2, all of the images are "stretched" inside
the viewer. Combinations of {alt|max}{height|width} do not seem
to alter this. Tapping on the image and selecting 'View image'
will show the image enlarged, and also stretched. A small
example of the situation is attached to this message.
6.) Using a maxdepth of 2 on a site which is very deeply linked to
offsite references (NewsBlaster again), tapping on a link that
is _NOT_ in the document will report a 'No URL information
found', as if I had used --no-urlinfo, but I did not (and it's
not in the ~/.pluckerrc file either). I have yet to produce a
pdb with this parser and viewer version which will show me the
actual URL of an out-of-range link in the pdb.
I think this is the result of the fix for the "Unhandled
Exception" errors that were happening with the previous parser
revision. Those no longer happen, but are now replaced with the
'No URL information' message.
I just spidered the same page down to a maxdepth of 3 (about
1,900 links, 200,000 references) and there was still a lack of
URL information, though it was now one link deeper before I
received the error than with maxdepth=2.
7.) When there is no url info present for a link, the dialog still
contains a "Copy URL" button. Tapping on this will then create
an entry in MemoPad with the time/date and a "No URL information
was found" message. Probably not ideal in the long run. If
there is no url information, we should lose the button that says
"Copy URL".
I'll keep testing it, but as I try to get some good sample content
to replace the stale "Samples" section on the current website, making sure
that content is usable becomes very important.
d.
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.0.7 (GNU/Linux)
iD8DBQE9GF6akRQERnB1rkoRAnfmAJwNv9uS/ak/rUJWTxDFYdJYkYBRdgCfbvk3
J0RnLvESTuDmBqO0Tah2Cbs=
=aDn8
-----END PGP SIGNATURE-----
nb-stretch.gif
Description: GIF image
