Mailing-List: contact derby-dev-help@db.apache.org; run by ezmlm
Precedence: bulk
Reply-To: <derby-dev@db.apache.org>
Received-SPF: unknown (idunn.apache.osuosl.org: domain pobox.com does not
 designate 64.74.244.70 as permitted sender)
Message-ID: <4508FE56.3070909@pobox.com>
Date: Thu, 14 Sep 2006 03:01:42 -0400
From: Geir Magnusson Jr <geir@pobox.com>
Reply-To: geir@pobox.com
User-Agent: Thunderbird 1.5.0.5 (Windows/20060719)
MIME-Version: 1.0
To: derby-dev@db.apache.org
Subject: Re: Why are you giving up???? (Was Re: Doc notice for JDBC 4
 functionality
 in Derby 10.2)
References: <45084177.8020701@bristowhill.com> <4508D53D.80406@pobox.com>
 <54ac72d70609132229k9039e10ia69991e825ed7ae5@mail.gmail.com>
In-Reply-To: <54ac72d70609132229k9039e10ia69991e825ed7ae5@mail.gmail.com>
Content-Type: text/plain; charset=ISO-8859-1
Content-Transfer-Encoding: 7bit


Andrew McIntyre wrote:
> On 9/13/06, Geir Magnusson Jr <geir@pobox.com> wrote:
>> Why are you giving up?
>>
>> I still believe there is a possible solution to this, so that the Derby
>> community can ship with JDBC4 capability - to that end, I'm doing what I
>> can to try to find a solution with Sun on this.
> 
> It wasn't clear to me, as I'm sure it was not clear to others, that
> anyone else was still pursuing a solution that would allow us to ship
> with the JDBC 4 bits in the binaries. Since the issue got stuck on the
> Mustang license with the Sun lawyers, and since the Sun people on the
> list seem to have abandoned the idea, I assumed the search for a
> solution was over. Thank you for continuing to pursue the issue.

I'm hope that I actually help here :)

> 
>> Do people not care?  I just don't understand.  Derby can be the world's
>> first database with JDBC4 support, so it's there and ready when Mustang
>> is released.
> 
> Unless there are some kind of major changes in between our release and
> the Mustang release that cause a major incompatibility on our side.
> Just recently, between b95 and b98 of Mustang, there was a few changes
> that caused major breakage. So if something similar happens between
> now and when Mustang ships, then we have the distinction of shipping
> the first database with really broken JDBC 4 support. I think this was
> Craig's concern. (not one of mine, necessarily, see below)

I understand.

> 
>> This means that Sun has to fork Derby, and  also JavaDB is therefore
>> more technically advanced than Derby, and no one wants that either. No
>> one wins here.  Lets find a solution.   I don't think it will take much
>> longer.
> 
> Even with the 'optional JDBC 4 functionality not built into the
> binaries' route for 10.2.1, there wouldn't be a need for Sun to fork
> Derby per se, they just wouldn't be shipping Apache's official
> release. They could still ship something mid-stream between 10.2.1 and
> 10.2.2 directly out of the Derby codebase with the JDBC 4
> functionality built in, and I personally wouldn't call that a fork.

That's true, and I guess I did get a little carried away there :)   I
was tired, in a plane, in the snow, at night, uphill, both ways...

> It's not clear to me that Sun was ever planning on shipping the
> official 10.2.1 anyway, since I'm pretty sure that Sun wanted to be
> up-to-the-second with the JDBC 4 spec and shipping the official
> release wouldn't let them do that. Can anyone from Sun clarify the
> plans for what would actually go into Mustang?

But how much will the API change towards the end of the spec vote?

Also, I would think that Sun would *want* to relabel a derby release,
because then support issues are much easier for the larger community to
deal with.

I think it would be much better for JavaDB to be in lock-step with
Derby, so a user of JavaDB would to be able to approach the Derby
community regarding questions about the code that could actually be
answered.  But this is Sun's call.

> 
> Anyway, I'd love to see Mustang ship with Derby, and for us to be able
> to ship 10.2.1 with JDBC 4 support in it sooner rather than later, so
> I'd love to hear the solution being pursued. Would the plan be for Sun
> to release the JDBC 4.0 API as a jar file under the spec license or
> some other compatible license so that we could use a 1.5 compiler to
> build in our JDBC 4 support? That seemed to be what you were
> suggesting in your the last mail.

Yes, and I've heard it's been considered and shot down.  I have some
other ideas - let me flesh them out a bit first.

> FTR, I don't find the compatibility concerns with 10.2.1 and Mustang
> terribly onerous, since we would have the JDBC 4.0 functionality
> clearly labelled as 'early and possibly not compatible with the final
> JDBC 4.0 spec,' or whatever language was being worked on, all over the
> docs and release notes. Plus, we could put out our own 10.2.2 with
> whatever changed and be up-to-spec the same week that Mustang is
> released.

Yes.

I was also wondering if this could be a plugin - that you drop the
derby-jdbc4.jar somewhere and it Just Works.  That artifact could be
released separately on the day of JDBC4 finality...

geir

> 
> cheers,
> andrew
> 
>