Thanks to Remko Troncon for the initial implementation; this is basically the
same with the logic simplified a bit.

This patch adds an optional "darcspath" parameter to the configuration,
allowing people with darcs in non-standard locations to specify it by simply
editing the config file.

Thanks to ale@bomberstudios.com who sent a patch to do this (although I did it
in a slightly different way).

This patch adds support for alternative patch views for all the different
patch displays (commitdiff, filediff, headdiff and headfilediff), supporting
currently a plain output and, for commitdiff, a raw one.

A darcs output is included but commented out, since I'm not familiar with
darcs' native output enough to make a syntax highlighter for it.

Besides, it also includes an alternative view for headblob (I was too lazy to
split up the patch =). Thanks to Zachary P. Landau who sent a patch to do this
(although I did it in a slightly different way).

If the repository was get with the --no-pristine-tree option, there won't be a
current directory; we only show the tree option if the directory is there.

The other related option that accesses the pristine tree directly is the
headblob, but there is already a check in place that takes care of it.

Thanks to Stephane Popinet for reporting it.

Some unicode stuff has changed in Python 2.4, so this change is needed to make
fixu8 work with older versions. It's not nice, but it does the trick and it
takes only a couple of self-contained lines.

Many people reported the problem and potential fixes, this one is by Markus
Keller <markus.keller@gmx.com>.

Skip decoding of an unicode string before encoding it in utf8, since it's not
supported by python 2.3

I have my doubts regarding this patch because I'm not sure removing the
decoding won't cause problems, but it went fine on my testing here using
latin1 and utf8 characters. I'd love to test it under a different encoding,
but I don't know any non-latin1 languages.

Implement support for repo names with semi-strange characters (ie. at least -
and _). It could use more improvement, but it'll do for most repos.

Skip decoding of an unicode string before encoding it in utf8, since it's not
supported by python 2.3

I have my doubts regarding this patch because I'm not sure removing the
decoding won't cause problems, but it went fine on my testing here using
latin1 and utf8 characters. I'd love to test it under a different encoding,
but I don't know any non-latin1 languages.