Tasks for navidoc 0.1alpha1: First separate release anybody: - umltool: mputils.dbg for handling screen output and logging - document our UML software (metacode/uml*, used by doc/uml/*) - reST -> latex (cleanup?) - fix umltool graphics to be closer to the UML 3amigos books - fix dep directive to be dep BuoyViewNode "calls" BuoyLinkListener instead of dep "calls" BuoyViewNode BuoyLinkListener currently. Involves fixing many diagrams! Reason: then can use in class: class BuoyViewNode dep "calls" BuoyLinkListener because inside class, the class is given as the 1st argument. - make pegboard to be only a directive, when all reST could be compiled using $(JYTHON) metacode/umldoc.py `find doc -name "*.rst"` - enhancing umltool - rationalize doc directory structure: - Gzz_Frontend_View.rst -> doc/frontend/View.rst (i.e. no Gzz_ prefixes, tree structure explicit) - nested classes in javadoc: grep for MipzipLoader.Level - a lot of things, which would be easier to handle with better design, so these will be fixed on the second implementation round with more object-oriented design Two of the following would be fixed with re-design of jlink class. - fix jlink focus matching problem, currently matching sees no difference e.g. between View (Java class) and Gzz_Frontend_Java.gen.html - write all errors and warnings in some place so it's easy to look for them and fix. Current output is incredibly bogously spewy. - warn properly about diagrams referencing classes that are not there. For instance, gzz.DimManager is referenced but does not exist. Maybe show jlink in some special color (bright red?) if false. - reST documents where diagram is explicitly included using the directive, should also be included into diagram (and highlighted when shown in current document) - make sure that umltool works also with pure doccxx - Currently diagrams are embedded into html-documents after the first header-tag. This could be enough for javadoc and other, but in reST this should be possible to overdrive by own directive. - diagram names should be unique, currently this can't be easily tested, make something for it - Fix UML sequence diagram: now you have to put seqobjects *after* the sequence... + Highlighting should be optional. This should be in UML source, but it could also be added there from an optin of UML directive. Anyway, implementing is not trivial, since even the same png diagram could be used in all documents, the imagemap should always be regenerated. So, two points: - all refers to the same diagram should use the same png - still every document needs own imagemap - there should be no highlighting - split this todo into correct releases / sections :) + clean pegboard.py (= make more oo, since Python is oo language) + check why running with jython causes harmless SyntaxWarning 1.0: Feature-complete anybody: - all UML diagrams in UML reference manual and user's guide should be easy to do - reimplement uml.py completely, or standardize on another UML tool; required features: - supports qualifiers, n-ary associations, class and sequence diagrams - supports arbitrary drawing on the diagrams - fast and easy to use and libre - figure out metapost tfm files; we need to have Helvetica.tfm since we want to use the postscript font names to get standalone files. But it would be nice not to have it in every directory ;)