Re: Gjdoc now runs on full classpath tree

[Julian Scheid]

Mark Wielaard wrote:
> I know of two things:
>   - Array method return types don't get the correct dimension.
>     See for example java.util.List.toArray().

This is now fixed in CVS.

>   - Sometimes the first sentence detection seems to fail.
>     See the summary of java.util.TimerTask.cancel().

Will check that.

> Speaking of java.util.TimerTask, should a code tag be treated as a pre
> tag? See the example code in the class description.

No, this doesn't work. You are supposed to use a pre tag, otherwise your
linefeeds will be ignored by the browser.

> And it would be nice to generate the documentation of each package in a
> separate directory. Currently a full run over Classpath gives you one
> big directory with 1350 html files.

The problem is that XSLT sheets can create new files in existing
directories, but can't create new directories. There are solutions
for this, but none is really nice. I'd prefer keeping the directory
layout for now... but I've added a task (#964).

> I looked at some other GNU projects that produce API documentation and
> we might want to see if we can handle those:
>
> http://www.gnu.org/software/kawa/api/
> http://www.gnu.org/software/classpathx/crypto/api/
> http://www.gnu.org/software/classpathx/jaxp/
>
> There are probably more. It would be nice if we could link them together
> without needing to get all source files and putting them in one
> directory. So you could have one file that lists the base URL of
> 'missing' classes/packages.

Yep right, I already had this on my mind. Added a task too, #965.

See you

Julian