As explained in IM, for the Javadocs there are a number of solutions.
But the best *overall* solution (I think) is to create stubs.  It
solves the problem of the release engineer needing access to
proprietary libraries also.

Karl


On Fri, May 25, 2012 at 5:41 AM, Simon Willnauer
<[email protected]> wrote:
> hey karl,
>
> maybe I misread this but can't your remote link the javadocs or 3rd party 
> libs?
>
> simon
>
> On Fri, May 25, 2012 at 11:16 AM, Karl Wright <[email protected]> wrote:
>> I think that would want to be a separate INFRA ticket.  We can't
>> really create it until our svn area is moved, and I'm still waiting on
>> that.
>>
>> Thanks,
>> Karl
>>
>> On Fri, May 25, 2012 at 2:35 AM, Alex Ott <[email protected]> wrote:
>>> Hi Karl
>>>
>>> Can we create git mirror for manifoldcf (as part of graduating
>>> activities)? I'm working on several computers and that would be handy
>>> to have "official" git repository instead of maintaining my own via
>>> git-svn
>>>
>>> On Mon, May 21, 2012 at 11:24 AM, Karl Wright <[email protected]> wrote:
>>>> Hi all,
>>>>
>>>> The Apache IT folks have an additional requirement which we will need
>>>> to fulfill at some point as part of the graduation process.  It is
>>>> apparently a requirement that at least the site be completely
>>>> buildable using only materials that we can obtain a zero-cost license
>>>> for.  We've been publishing our site including javadoc including the
>>>> documentum, filenet, livelink, and meridio connectors, which
>>>> technically violates this principle.
>>>>
>>>> There are two ways to address the problem.  The first way is to check
>>>> into trunk somewhere the javadocs, and periodically regenerate/check
>>>> in changes.  The second way is to write stubs for all of the
>>>> proprietary libraries, and build and javadoc against those stubs
>>>> rather than against the actual libraries themselves.
>>>>
>>>> After thinking about this a while, I concluded that it might be
>>>> feasible to create stubs, and there would be additional benefits to
>>>> the approach.  Specifically, it would no longer be necessary for our
>>>> release engineer to obtain the necessary proprietary materials in
>>>> order to produce a release, and that would be a huge benefit in the
>>>> long run.  Also, Apache IT communicated that they were concerned that
>>>> some of the proprietary materials might have javadoc restrictions such
>>>> that we could not legally release javadoc for code that compiled
>>>> against these jars.  I do not think this is true but in the interests
>>>> of trying to be a good Apache citizen we should eliminate this concern
>>>> if it is reasonably possible to do so.
>>>>
>>>> I've therefore created a ticket, CONNECTORS-474, and a corresponding
>>>> branch, and am working away on this.  It will, of course, take some
>>>> time.  There will also need to be some testing of the proprietary
>>>> connectors needed before the branch is committed, because building
>>>> against stubs might inadvertantly introduce linkage errors.  In the
>>>> interim I've been assured that the migration to our TLP resources will
>>>> continue as planned.
>>>>
>>>> Thanks,
>>>> Karl
>>>
>>>
>>>
>>> --
>>> With best wishes,                    Alex Ott
>>> http://alexott.net/
>>> Tiwtter: alexott_en (English), alexott (Russian)
>>> Skype: alex.ott

Reply via email to