Agreed with #1
Let me better explain #3. It is not a taglet copying Sun's spec. Instead
it should insert pointers to specification of actual methods on Sun's
website. So the resulting docs would look like the following:
public Foo(type arg)
Spec reference: See corresponding <a [link to Sun's spec]>J2SE API
specification reference</a>
And some additions, like "The spec allows this and that behavior, we
did it this way because ..."
What IMHO would not be a good idea is rewording complex method
descriptions as we could unintentionally change semantics and mislead
people who read our docs rather then official spec.
Thanks,
Mikhail Loenko
Intel Middleware Products Division
-----Original Message-----
From: Tim Ellison [mailto:[EMAIL PROTECTED]
Sent: Tuesday, January 10, 2006 4:26 PM
To: harmony-dev@incubator.apache.org
Subject: Writing JavaDoc (was: Re: [RESULT] ( Was Re: [VOTE] Accept
JIRA contribution HARMONY-16
(Intel's contrib of security code for classlib)))
Loenko, Mikhail Y wrote:
Thanks for accepting the contribution
There's a bit of things that come out of this, like the
"com.intel.drl.spec_ref" javadoc tag that we should convert, and
such.
What would be the best for those javadocs? We can have 3 possible
options:
1. Copy-paste from the spec. Not sure it is legal
This one definitely has to be out. The Sun JavaDoc is a
copyrighted/licensed work so making a verbatim copy is unacceptable.
2. Reword the spec. More likely to be legal
As I see it, the JavaDoc fulfils (at least) two purposes. It embodies
the java spec (i.e. the definition of the standard library's
behaviour),
and it is the principal developer documentation (i.e. how to use the
library). We do not want to change the specification in any way, but
can enhance the usability of the documentation to developers.
For example, it would IMHO be wrong to specify the behaviour of a
method
with more/less restrictions than the original reference javadoc,
because
that implies that developers can make assumptions on one implementation
that they cannot on the other. However, it is reasonable to give more
examples, usecases, even performance, threading guidelines, etc. that
do
not change the functional specification.
So I'd say writing some JavaDoc, that was neither a direct copy of the
original, nor 'enhancing' the specification, can provide value to
developers.
3. Replace the tag with a different one and provide taglet to build
the
doc from the Harmony sources and Sun's spec.
If I understand this correctly, then I don't see how this is
substantially different to option (1)? Whether it is a human that does
the cut-n-paste into the Harmony release, or a doclet, the result
includes somebody else's work.
Regards,
Tim
Currently IBM's contribution seems to have #2. Does anyone have an
opinion?
Thanks,
Mikhail Loenko
Intel Middleware Products Division
-----Original Message-----
From: Geir Magnusson Jr [mailto:[EMAIL PROTECTED]
Sent: Friday, December 30, 2005 10:48 PM
To: harmony-dev@incubator.apache.org
Subject: Re: [RESULT] ( Was Re: [VOTE] Accept JIRA contribution
HARMONY-16 (Intel's contrib of
security code for classlib))
Tim Ellison wrote:
Geir Magnusson Jr. wrote:
<snip>
I'll finish moving to SVN and we'll put into the classlib tree, I
suppose.
Great. Thanks again Mikhail and the team!
I suggest you either put it into the classlib tree at
"classlib/java-src/security2" or leave it in the sandbox, then we
can
merge it into the existing security structure without breaking the
world.
I'll go for the former and try to whip it into common shape, and we
can
then decide how we do this - drop the existing security if security2
is
a superset, or merge.
There's a bit of things that come out of this, like the
"com.intel.drl.spec_ref" javadoc tag that we should convert, and
such.
Also will give me a good change to frame out the test infra.
Regards,
Tim
--
Tim Ellison ([EMAIL PROTECTED])
IBM Java technology centre, UK.