Description of problem: On the PPC platform, LaTeX2HTML does not generate included images (e.g. for mathematical expressions), as the image processing toolchain breaks. The main bug appears to be in /usr/bin/pnmcolormap, which is called from the /usr/bin/pstoimg script. In addition, /usr/bin/pnmtopng segfaults on empty input. This is a secondary bug triggered by the first. Version-Release number of selected component (if applicable): latex2html-2002.2.1-3 How reproducible: Always. Steps to Reproduce: Overall bug: Run latex2html on any file containing math and/or images. Result: math/images will not show in the generated HTML. Bug in /usr/bin/pnmcolormap: Is exposed by running $ pnmcolormap 256 -verbose /tmp/p3203.pnm pnmcolormap: making histogram... pnmcolormap: Scanning image 0 pnmcolormap: 2 colors so far pnmcolormap: Scanning image 1 pnmcolormap: Image 1 maxval (1) is not the same as Image 0 (255) Where /tmp/p3203.pnm is one of the intermediary file generated by latex2html. Secondary bug in /usr/bin/pnmtopng: $ /usr/bin/pnmtopng < /dev/zero Segmentation fault Actual results: see above Expected results: Same commands on an i386 system (this is Suse 9.0, as I currently don't have access to a fedora i386 machine): > pnmcolormap 256 -verbose /tmp/p10758.pnm pnmcolormap: making histogram... pnmcolormap: 2 colors found pnmcolormap: Image already has few enough colors (<=256). Keeping same colors. pnmcolormap: Generating 2 x 1 image P5 2 1 255 > /usr/bin/pnmtopng < /dev/zero [hangs] Additional info:
Netpbm is at fault here, so changing the component. I fixed the latest pnmtopng in netpbm-10.33 what caused some problems. Which netpbm do you use? Could you please test it with the latest rawhide netpbm (10.33-2)? I tried that and latex2html works for me with a sample tex file with included image. I release FC updates when the fix is confirmed.
I was using netpbm-10.33 on a fully up-to-date fc4. The binary netpbm from rawhide does not install on fc4, so compiled the source RPM. Lot's of compiler warnings, looks scary... In any case, my LaTeX2HTML runs now work for me - thanks!
Yes, the warnings need some upstream care to not to look that scary ;) Many of them might be a source of another bugs caused mostly by usage of uninitialized variables. *** This bug has been marked as a duplicate of 188594 ***