additional errors happen in the last step.  The trick is to avoid
division by 0.1**n -- multiply by 10.0**n instead.

  - When facility not specified to syslog() method, use default from openlog()
    (This is how it was claimed to work in the documentation)
  - Potential resource leak of o_ident, now cleaned up in closelog()
  - Minor comment accuracy fix.

the comments.

method='GET', always get the query string from environ['QUERY_STRING']
or sys.argv[1] -- ignore an explicitly passed in fp.

Avoid extra level of recursion in the paper-*/ directories.

Added the "info" target to use the subdir.  Partparse, however, is broken
again.

	"big" targets.  Fix is to set the TEXINPUTS on the command line of the
	sub-make.

the name of the main .tex source file if it's not in the current directory.

	specifying *which* type of path to search.  This works for both
	teTeX 0.4 and 0.9.  This is what we call portability these days!  ;-)

file; "make" (the 'all' target) became a no-op!

Now works as previously described.

Added some new cruft.

For some of the "big" targets, force things to happen in the "right"
subdirectory, i.e., "make" will build the DVI and PostScript files in the
paper-letter/ directory, and "make PAPER=a4" will build DVI and PostScript
files in the paper-a4/ directory.

Probably still needs some restructuring.

most of the processing is the same and has to be done anyway.

Using \input instead of \include to avoid extra .aux files from being
generated.

Some targets may be a little raw, but the basic formatting targets are all
tested, primarily for use from the subdirs.  There are probably a few
gotchas, and the paper-*/ directories could use "helper" Makefiles.

Most of the logic is still in the primary Makefile and a new script in
Doc/tools/.

enough platforms).