On Jan 7, 2006, at 12:05 PM, Ian Bicking wrote:
Bob Ippolito wrote:
Probably mostly broken patch to get the source code to build for
modules that don't show up in __all__ (so the JSONEncoder and
JSONDecoder source links work):
http://lesscode.org/projects/pudge/ticket/8
The patch seems fine. At least it's easy to understand ;) I want
to look through how the finding works, as I think there needs to
some more ways to indicate things:
* Some modules should not be listed. I have backported stdlib
modules for instance, and I don't want these to show up in the
documentation.
* Some modules should not show up in the module index. Deprecated
modules, for instance.
* Some modules appear to belong to the package, but don't really.
I'm having this problem with namespace packages in paste (multiple
projects use the paste package).
* Some objects should not be displayed in the module index. This
includes deprecated objects, and private objects.
* Some private objects should still have their docstrings rendered,
because there interface is public in some way. For instance, when
a function returns an instance of some class -- the class itself
may be private, but the interface of the class is public. I think
this should be handled by putting in a link in the function
docstring to the class, but the class itself shouldn't be in the
index.
* Some objects are fully public, but don't show up in __all__. I
added a __pudge_all__ variable that overrides __all__ just for the
purposes of pudge for this.
So I think some distinction more than just isvisible is called for.
I agree.. but I had a simple problem to solve so I just wrote the
minimal amount of code that it took to make that happen... though it
did take a little too long to figure out how to make that happen.
Relatively big patch to use "unobtrusive javascript" techniques
to make the docs work without a webserver by stowing away the
line range inside of a class name rather than directly on the
query string:
http://lesscode.org/projects/pudge/ticket/9
Is this a Safari issue or somesuch? Firefox works fine for me in
local browsing.
Probably
I think this patch is against an older version; I updated the
Javascript recently to do better highlighting. But I haven't
actually tried applying it. Ryan can give you commit access too if
you want it.
It was definitely against the trunk when I wrote it about a week
ago :) Commit access would be fine, not promising I'd use it much
but it'd make things a little easier for me if I find an issue.
I was actually thinking that simplejson might have a better home
under the lesscode umbrella too.
I'm also just copying over one of the files on top of index.html
with this script, because the module index is all I really needed:
http://svn.red-bean.com/bob/simplejson/trunk/scripts/make_docs.py
I find the index handling funny in pudge; I think index files could
be treated just like normal files, and then this shouldn't be a
problem.
In this case I wanted the module index as the index page, not some
separate index file.
In the future, it would be nice if more license templates were in
there. Personally I'd like to use either a public domain or CC
attribution license. No patch on this one:
http://lesscode.org/projects/pudge/ticket/6
Do we need to include the text of the license on the site itself?
Is a link to the original license sufficient? If so, I'd like to
make the license field at least optionally a link.
I don't know, I was just following the convention of the two included
licenses. Seems strange to only provide two choices, both of which
kinda have the same implications.
-bob
_______________________________________________
Pudge mailing list
[email protected]
http://lesscode.org/mailman/listinfo/pudge
