Text in the right margin determines the milestone for the feature (R1
means release 1, i.e. version 1.1).

TODO:

 o correct startTagForNode(node) to generate correct XML (watch out on    [R1]
   quoting, entities, etc.)
   Perhaps I should use node.serialize, and try to get only the stuff
   between first "<" and ">"?
   [This is a BUG, needs to be sorted out soon]

 o When outputting messages to PO file, we need to escape "\" as well.    [R1]

 o add ability to merge translator-credits back in the XML file           [R1]
   (I figure this would be best done via uncommenting a region and
   replacing translators names there)

 o install xml2po in $prefix, data in $prefix/share, and document         [R1]
   descriptions in $prefix/share/xml2po/docbook/*, 
   $prefix/share/xml2po/summary/*, etc.

 o allow tags which preserve spaces                                       [R1]
   (for now, you certainly want all preformatted tags in ignored-tags)

 o add support for setting current language as either attribute of        [R1]
   some element, as another child element of existing element
   (eg. DocBook articles use "<article lang='lang'>", where lang is
   basename of the PO file without ".po" (for sr.po, it's simply "sr").

 o add support for translating attributes (eg. imagine "title"            [??]
   attribute in IMG tag in XHTML) -- this should be easy enough

 o make "descriptions" of document types which contain lists of final     [R2]
   tags, ignored tags, attributes to be translated, etc. install them
   in $prefix/share/xml2po/ (this will deprecate options -i and -f).

 o make "-p POFILE" read directly PO file, instead of depending on msgfmt [R2]

BUGS:

 No (undocumented :) bugs known.

DONE:

 o add "-a" option to automatically detect final tags (should be          [??]
   simple, use something like worthOutputting in isFinalTag)

 o fix normalizeString not to assume that inserting spaces is ok
   [This should be good enough now]

 o allow nesting of tags in final-tags which are then replaced by
   something like "<tag id='blah'>"

 o source code references are working now

 o Implement -o option for directing output into file.

 o Recognize comments before tags as translators' comments.

 o Add option "-p POFILE" which will run msgfmt for user automatically.

 o Add option "-u LANG" which works like "intltool-update LANG"
   (runs msgmerge for the user and works on file LANG.po)

 o Entities which contain tags are not handled correctly, this can be 
   easily solved using ctxt.replaceEntities(1) -- that's how it was done

 o Ignore external entities (first need to look if it's at all possible)
   This seems be impossible with Python bindings, C interface provides
   xmlEntity.etype field (enum xmlEntityType), but this doesn't seem to 
   be available from Python bindings.  Because of this, I hacked
   around it using xmlNode.debugDumpNode to a temporary file, what
   could sure use some improvement.
