Mailing-List: contact dev-help@geronimo.apache.org; run by ezmlm
Precedence: bulk
Reply-To: dev@geronimo.apache.org
Received-SPF: pass (athena.apache.org: local policy)
DomainKey-Signature: a=rsa-sha1; q=dns; c=nofws;
  s=s1024; d=yahoo.com;
  h=Received:X-Yahoo-SMTP:X-YMail-OSG:X-Yahoo-Newman-Property:Content-Type:Mime-Version:Subject:From:In-Reply-To:Date:Content-Transfer-Encoding:Message-Id:References:To:X-Mailer;
  b=iysEcVs6aGEZSz7ZkcVCmJSAjTAUmwBFeo6G1b/umaeMyZUgjM8IoN3cU9eR+0a0vKuHi1w7WTCY4i+pF4Din/Jw12/zXJ2yh+nh2m4y+gTsMVXx+xzxGT6kRUxyxSdRAJYOfyrpHd0plZwvI0amtCfokFjvXxvckbJp5DTKSBQ=
  ;
Content-Type: text/plain; charset=us-ascii
Mime-Version: 1.0 (Apple Message framework v1077)
Subject: Re: Spec release numbers
From: David Jencks <david_jencks@yahoo.com>
In-Reply-To: <4BBB4B08.3060809@gmail.com>
Date: Tue, 6 Apr 2010 11:23:16 -0700
Content-Transfer-Encoding: quoted-printable
Message-Id: <ABA5C6D9-440C-42E3-BF91-6C85A87DF6DB@yahoo.com>
References: <4BBB3FB1.40500@gmail.com> <4BBB467D.60508@apache.org>
 <4BBB4B08.3060809@gmail.com>
To: dev@geronimo.apache.org


On Apr 6, 2010, at 7:54 AM, Rick McGuire wrote:

> On 4/6/2010 10:34 AM, Donald Woods wrote:
>> Should we do like the server releases?
>> The first new major release uses 2 digits and any follow-on =
maintenance
>> releases introduce the third digit, like -
>>   2.1, then 2.1.1, 2.1.2, 2.1.3, ....
>>  =20
>=20
> That's certainly not what's been done.  Going through the numbers, we =
have a mixture of 1.0-SNAPSHOTS and 1.0.0-SNAPSHOTS for first releases, =
and also a mixture of maintenance release numbers with both 2 and 3 =
digits.  Your suggestion sounds fine to me, though the specs really =
never use the middle digit...or even the first one, for that matter!.
>=20
>> Bigger question, is what does OSGi want?  When we set version ranges
>> like [1.0,2.0) does having 1.0 vs. 1.0.1 artifacts matter?
>>  =20
>=20
> The bundles don't export the version numbers of the spec =
implementation, but only the version number of the interface they =
implement.  Thus the 1.0.0 version of geronimo-el_2.2_spec exports the =
2.2 version of the el packages.  The 1.0.0 version number does not show =
up there.

IIUC osgi expects to come up with some package versions for ee spec apis =
that has no relationship to the ee spec version but does follow osgi =
version semantics.  So, anything we publish now will be at least 90% =
wrong and apt to produce problems if it's released into the wild for =
everyone.  The only case I'm aware of where there's been some conclusion =
is to version the jpa 2 api at 1.1.

Maybe we should not version the packages in the jars until osgi firms up =
its idea of what they should be????  =20

thanks
david jencks

>=20
> Rick
>>=20
>> -Donald
>>=20
>>=20
>> On 4/6/10 10:05 AM, Rick McGuire wrote:
>>  =20
>>> I've been going through and doing some release dry runs on the spec
>>> projects, and I've noticed that there is an inconsistency with the
>>> release numbering.  Some of the projects use a two level release =
number
>>> (e.g., 1.0), while others use a three level numbering system (e.g.,
>>> 1.0.0).  It would be nice to make these consistent, and since we're
>>> going to be releasing most of these shortly, now seems like a good =
time
>>> to do this.
>>>=20
>>> So, the question I have is which system should we use?  Many =
projects
>>> use a 3-level system, but in the case of the specs, I don't believe =
we
>>> ever really use the middle digit when 3 levels are used.  That =
generally
>>> would only occur when there are functional enhancements to the spec,
>>> which generally results in a new subproject getting created to =
reflect
>>> the spec number change.
>>>=20
>>> So, should we:
>>>=20
>>> [] Convert everything to two-digits
>>> [] Convert everything to three-digits
>>> [] Leave things the way they are
>>>=20
>>> If the consensus is we're fine the way we are, then the next =
question is
>>> whether we should be using two or three digits for newly created =
spec
>>> projects.
>>>=20
>>> Rick
>>>=20
>>>    =20
>>  =20
>=20