Mailing-List: contact dev-help@felix.apache.org; run by ezmlm
Precedence: bulk
Reply-To: dev@felix.apache.org
Received-SPF: neutral (nike.apache.org: local policy)
MIME-Version: 1.0
In-Reply-To: <4C5C0E97.7010809@ungoverned.org>
References: <AANLkTimc5ZiWhkGMRriogWG7f506--L2nfScB-mHdHk+@mail.gmail.com>
	<4C5ABA50.9050907@ungoverned.org>
	<AANLkTi=simZ_W+97TtB4xis2VjP=wJg3BWR38p6WeF0H@mail.gmail.com>
	<4C5B02FF.1080305@ungoverned.org>
	<AANLkTikZdyt0Oq70t5jgYpOztEJdKSQaVPyOtFfD5JCg@mail.gmail.com>
	<4C5B411D.5090608@ungoverned.org>
	<AANLkTimQuskB12yfX3X-C42Er86+-TW9JxXes2QqYx6q@mail.gmail.com>
	<4C5C0E97.7010809@ungoverned.org>
Date: Fri, 6 Aug 2010 15:39:14 +0100
Message-ID: <AANLkTi=GZiv=BKLM7umNTSaSrJCW5vf0KCF2Ad84AyP9@mail.gmail.com>
Subject: Re: A long time ago...
From: David Savage <david.savage@paremus.com>
To: dev@felix.apache.org
Content-Type: text/plain; charset=ISO-8859-1
Content-Transfer-Encoding: quoted-printable

On Fri, Aug 6, 2010 at 2:31 PM, Richard S. Hall <heavy@ungoverned.org> wrot=
e:
>
>
> On 8/6/10 6:33, David Savage wrote:
>>
>> On Thu, Aug 5, 2010 at 11:54 PM, Richard S. Hall<heavy@ungoverned.org>
>> =A0wrote:
>>>
>>> On 8/5/10 5:43 PM, David Savage wrote:
>>>>
>>>> On Thu, Aug 5, 2010 at 7:29 PM, Richard S. Hall<heavy@ungoverned.org>
>>>> =A0wrote:
>>>>>
>>>>> =A0On 8/5/10 11:47, David Savage wrote:
>>>>>>
>>>>>> On Thu, Aug 5, 2010 at 2:19 PM, Richard S. Hall<heavy@ungoverned.org=
>
>>>>>> =A0wrote:
>>>>>>>
>>>>>>> =A0On 8/5/10 6:41, David Savage wrote:
>>>>>>>>
>>>>>>>> Hi there,
>>>>>>>>
>>>>>>>> I realise it's been quite a while since we donated Sigil to Apache
>>>>>>>> and
>>>>>>>> I'm yet to push out a release. That said I've been making quite a
>>>>>>>> bit
>>>>>>>> of progress with it in the background [1] and I'd like to start
>>>>>>>> figuring out what tasks I need to do to get these bundles released=
.
>>>>>>>>
>>>>>>>> Signing jars seems to be one task that needs doing, also setting u=
p
>>>>>>>> appropriate LICENSE files, but I'm sure there's other stuff. Havin=
g
>>>>>>>> not pushed out an apache release before I'd appreciate any pointer=
s
>>>>>>>> to
>>>>>>>> get me going.
>>>>>>>
>>>>>>> The main things are:
>>>>>>>
>>>>>>> =A0 * Make sure that all files of any significance have the Apache
>>>>>>> =A0 =A0 header in them.
>>>>>>> =A0 * In the root of all bundle projects, include LICENSE, NOTICE, =
and
>>>>>>> =A0 =A0 DEPENDENCIES files.
>>>>>>> =A0 =A0 =A0 =A0 o LICENSE is the standard license text, NOTICE cont=
ains any
>>>>>>> =A0 =A0 =A0 =A0 =A0 required notices from included software, and DE=
PENDENCIES
>>>>>>> is
>>>>>>> =A0 =A0 =A0 =A0 =A0 like an expanded NOTICE where we acknowledge al=
l top-level
>>>>>>> =A0 =A0 =A0 =A0 =A0 dependencies.
>>>>>>> =A0 =A0 =A0 =A0 o These files should ultimately also end up in the =
META-INF/
>>>>>>> =A0 =A0 =A0 =A0 =A0 directory of the resulting bundle JAR file.
>>>>>>
>>>>>> Ok makes sense - just to be clarify I've setup the sigil projects
>>>>>> under the following structure:
>>>>>>
>>>>>> $sigil/common - has dependencies on bnd and osgi.framework and
>>>>>> osgi.cmpn
>>>>>> $sigil/ivy - has dependency on ivy, common + common deps
>>>>>> $sigil/eclipse + has dependency on eclipse, common + common deps
>>>>>>
>>>>>> Guess it make sense to have different NOTICE/DEPS for each sub modul=
e?
>>>>>
>>>>> If you are planning to only have a single combined release of
>>>>> everything
>>>>> (i.e., every release is monolithic), then you can just have one set f=
or
>>>>> all
>>>>> of them. However, I'd imagine you'd keep everything modular and allow
>>>>> for
>>>>> different release schedules for the various modules, if so then you
>>>>> need
>>>>> one
>>>>> set per module.
>>>>
>>>> Right I am debating doing a common/ivy release first as that stuff is
>>>> pretty rock solid, the eclipse subsystem is very usable but if it was
>>>> a perfect world I'd finish off the runtime/debug support before
>>>> pushing that to 1.0
>>>>
>>>> There are sub bundles within those top level subsystems but in general
>>>> I think it makes sense initially to release them as a unit, possibly
>>>> subsequent bug fixes can be done individually...
>>>>
>>>> So yep looks like I need files per subsystem (at least) at the moment.
>>>
>>> If you think there's a chance to release them independently, I'd just g=
o
>>> ahead and create the files now if I were you, since it isn't that much
>>> work.
>>
>> Ok I'll give it a go, other thought that just occurred - I assume I
>> should tag releases in svn - do you usually tag release candidates too
>> - probably overkill?
>
> Only the release ends up with a tag.
>
>> Looking at the felix releases tags in svn [1] I guess I should tag all
>> the sigil bundles individually too - so that I can do individual
>> releases later.
>
> Think of it this way: one release tag for each individually downloadable
> unit. If this time around you are going to only package everything up in =
one
> tar ball, then you only need one. You can change that in the future.
>
> Of course, there is sort of an assumption that everything being released =
now
> has the same version number. If you break things out later, it would be o=
dd
> if individual bundles had a lower version number than this first lumped
> together release.

Right, everything in the sigil bundles is at 0.9.0 at the moment and
when I push the initial release I'll bump it forward to 1.0.0

In effect there are two deliverables from the sigil build at the
moment - though they depend on the various underlying bundles.

There's the eclipse update site - which contains a bunch of bundles
(as you'd imagine) and there's the
org.apache.felix.sigil.ivy.resolver.jar which is a plugin module for
ivy and aggregates all the sigil.common and sigil.ivy packages into
one bundle for convenient install in ant.

To build the ivy plugin, depends on sigil.common and sigil.ivy and to
build the eclipse update site depends on sigil.common and
sigil.eclipse...

So it seems like one way to go would be to tag common and ivy as 1.0
first. Then when the eclipse integration is ready for a 1.0 (retag
common as 1.2 - if it's changed) and tag eclipse as 1.0.

Of course I might also have to do some jiggling of the build scripts
for this to work as I guess we need to be able to build the sub
modules individually? - currently common, ivy and eclipse depend on a
shared bldcommon directory to specify global config - potentially
bldcommon should be tagged as well.

So the final steps to build the release would be to, tag, tag, tag etc

svn co .../releases/sigil.bldcommon-1.0 bldcommon
svn co .../releases/sigil.common-1.0 common
svn co .../releases/sigil.ivy-1.0 ivy

ant -f bldcommon/build.xml clean dist

Or something similar...??

Regards,

Dave

>
> And like Stuart says, the source release is the real important one, but
> typically we provide a convenience binary build for each source release w=
e
> do.
>
> -> richard
>
>
>
>> Regards,
>>
>> Dave
>>
>> [1] https://svn.apache.org/repos/asf/felix/releases/
>>
>>> -> =A0richard
>>>
>>>> Regards,
>>>>
>>>> Dave
>>>>
>>>>> So, I guess you need to answer that question first.
>>>>>
>>>>> -> =A0 =A0richard
>>>>>
>>>>>>> =A0 * Then just follow the release steps in our development
>>>>>>> =A0 =A0 documentation section for Nexus, which discusses signing, e=
tc.
>>>>>>
>>>>>> Thx I'll take a read through.
>>>>>>
>>>>>>> That's pretty much it, I think. You can look at other subprojects f=
or
>>>>>>> specific examples or just ask.
>>>>>>
>>>>>> Great, will do.
>>>>>>
>>>>>> In terms of staging release artifacts should I push these to my
>>>>>> people.apache.org/dsavage dir - or is there a folder I can access fo=
r
>>>>>> felix?
>>>>>>
>>>>>> Regards,
>>>>>>
>>>>>> Dave
>>>>>>
>>>>>>> In the end, you don't have to worry too much, because it's an
>>>>>>> iterative
>>>>>>> process when you call the vote...we'll review the release then, whi=
ch
>>>>>>> may
>>>>>>> cause you to have to re-do it.
>>>>>>>
>>>>>>> -> =A0 =A0 =A0richard
>>>>>>>
>>>>>>>> Regards,
>>>>>>>>
>>>>>>>> Dave
>>>>>>>>
>>>>>>>> [1]
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>> https://issues.apache.org/jira/secure/IssueNavigator.jspa?reset=3D=
true&&pid=3D12310100&fixfor=3D12314109&sorter/field=3Dissuekey&sorter/order=
=3DDESC&sorter/field=3Dstatus&sorter/order=3DASC
>