word2x developer's corner

If you hack code in your sleep, think spending lots of time using GDB is good fun, and have a machine a tad higher spec than mine :-), then word2x development can absorb all your spare time and more. (An aging 486DX2/50 does not help doing a quck recompile before it dumps core again).

Links

• Return to the Home page
• Public information about the word binary format is here. No information affected by non-disclosure agreements is supplied.
• Source file breakdown
• tok_trans API documentation.

How do I get the latest development code?

• Removing the cerr<< stuff in reader/reader.cc seems to cause core dumps in the dequeue operation when reader/mutliplex.cc calls the read() method. Any brilliant ideas why? (glibc 2.0.6, libg++ 2.8.1, bunutils 2.8.1 and g++ 2.8.1 on my system).
• reader/eqnarr.cc and reader/sections.cc need finishing.
• The LaTeX and HTML output code needs updatuing
• Lots of Makefiles need updating

Please note this list may be out of date due to other things getting in the way and developers fixing the problems :-)

New features

The next release is completely resdesigned due to the completely dire choas that was affecting the previous release. Word97's unicode document were rasing their menacing head leading to the need to for more support. Thus some internal infrastructure and major code cleanup was required. Already done are:

• Junk stripping code that understands unicode documents.
• Modular reader stages that can be combined flexibly.
• Modular command line option processing.
• The ability to tune all the tunable stuff from the command line.
• Ability to specify custome processing on the command line.
• A module that prints the passing tokens.
• A module that queues tokens and forwards them in large batches (useful in conjunction with the module above).
• Better table processing. However it could still be improved.
• Some programming documentation.

As time progresses word2x increasingly simulates the resource hog nature of word. I think the overall problem is AI-complete :-)

Further jobs

If you feel unable to do any of the jobs above, here is the rest of the endless list of jobs you might like to try. It is not exhaustive or even close.

• libole needs upgarding to a usable state
• Some binary junk undersatnding readers would be nice for accurate conversion.
• Backend functionality that would benefit all backends should be changed into reader stage(s). IT is not cheating to capture something in more that one stage (the new table code is 2 , or arguably 3).
• wmf to eps, bmp to pnm, and MS eqaution to *TeX code. Finish libole before starting these.
• *roff fancy ascii text output.
• Maybe *roff source output.
• Support for using getopt instead of getopt_long, at a cost to functionality (partial).
• Fix all the bugs.
• Update the makefiles.
• Foolproof operation.
• Write an RTF understanding input stage.
• Applixware input and output for those that use that package.