[svn r38388] Removed stuff that's done, added a couple more items to do.

--HG--
branch : trunk
This commit is contained in:
guido 2007-02-10 15:22:16 +01:00
parent ada3a4cfcf
commit f88bbf1b83
1 changed files with 2 additions and 82 deletions

View File

@ -1,23 +1,4 @@
* format docstrings more nicely (with tests)
DONE I guess
* have the API function view be as informative as possible
without having to go to the "single method" view
(do we even need a single method view?), for example:
browsing the class views (and clicking on methods)
should always make it obvious which class is being
viewed. method views (when navigating there through
the class view) should also have the source there
DONE I think, single method view is gone
* have class-level attributes be displayed
DONE
* use "inherited" doc strings, i.e. for * use "inherited" doc strings, i.e. for
class A: class A:
def meth(self): def meth(self):
@ -31,68 +12,7 @@
NOT YET DONE (later?) NOT YET DONE (later?)
* factor out some common code in the build_* functions * add SVN info to the (source only?) pages
(mostly) DONE * add warning about py.test possibly not covering the whole API
* refactor the apigen/rsession interaction to become
cleaner (e.g. apigen's get_documentable_items should
be separately tested and the caller should not need
to guess what it will get, i think)
DONE
* look out for and streamline all apigen/source-viewer
documentation into one document
* consider automating dependencies:
e.g. something like: queue_render(page, fspath, linker, ...)
would defer the rendering until later.
then a loop does:
maxlength = len(queue)
while queue:
page, fspath, linker, ... = queue.get()
# fill outputpath/link here or even earlier
if all_links_resolve(page, linker):
render it and write to filesystem
maxlength = len(queue)
else:
queue.append(...)
maxlength -= 1
if maxlength <= 0:
print "ERROR: seems i can't make progress"
print "unresolved links follow: "
...
print "unresolved pages/fspaths:"
...
XXX maybe: print "filling linker with dummy hrefs, and rendering anyway"
...
raise ...
NOT SURE if this is still required
* also we might have a support function for tests that
fills the linker with "dummy hrefs" for certain types
like source links
KIND OF DONE, the tests now use a linker that just doesn't
barf on non-existing linkids anymore, which seems to be
good enough (we may want to add more sophisticated debugging
later, but for now this works)
* add syntax coloring for Python source snippets
DONE
* remove py.test/apigen cruft from stack traces
DONE, thanks to fijal
* fix non-ascii source encoding support
DONE
* XXX