----- Original Message -----
From: "Juan Hernandez" <jhernand(a)redhat.com>
To: "Alon Bar-Lev" <alonbl(a)redhat.com>
Cc: engine-devel(a)ovirt.org
Sent: Wednesday, January 30, 2013 7:44:42 PM
Subject: Re: [Engine-devel] Time to move to 3.3.0-SNAPSHOT?
On 01/30/2013 06:22 PM, Alon Bar-Lev wrote:
>
>
> ----- Original Message -----
>> From: "Juan Hernandez" <jhernand(a)redhat.com>
>> To: "Alon Bar-Lev" <alonbl(a)redhat.com>
>> Cc: engine-devel(a)ovirt.org
>> Sent: Wednesday, January 30, 2013 7:17:11 PM
>> Subject: Re: [Engine-devel] Time to move to 3.3.0-SNAPSHOT?
>>
>> On 01/30/2013 05:58 PM, Alon Bar-Lev wrote:
>>>
>>>
>>> ----- Original Message -----
>>>> From: "Juan Hernandez" <jhernand(a)redhat.com>
>>>> To: "Alon Bar-Lev" <alonbl(a)redhat.com>
>>>> Cc: engine-devel(a)ovirt.org
>>>> Sent: Wednesday, January 30, 2013 6:49:10 PM
>>>> Subject: Re: [Engine-devel] Time to move to 3.3.0-SNAPSHOT?
>>>>
>>>> On 01/30/2013 05:33 PM, Alon Bar-Lev wrote:
>>>>>
>>>>>
>>>>> ----- Original Message -----
>>>>>> From: "Juan Hernandez" <jhernand(a)redhat.com>
>>>>>> To: "Alon Bar-Lev" <alonbl(a)redhat.com>
>>>>>> Cc: engine-devel(a)ovirt.org
>>>>>> Sent: Wednesday, January 30, 2013 6:30:55 PM
>>>>>> Subject: Re: [Engine-devel] Time to move to 3.3.0-SNAPSHOT?
>>>>>>
>>>>>> On 01/30/2013 05:13 PM, Alon Bar-Lev wrote:
>>>>>>>
>>>>>>>
>>>>>>> ----- Original Message -----
>>>>>>>> From: "Juan Hernandez"
<jhernand(a)redhat.com>
>>>>>>>> To: "Alon Bar-Lev" <alonbl(a)redhat.com>
>>>>>>>> Cc: engine-devel(a)ovirt.org
>>>>>>>> Sent: Wednesday, January 30, 2013 5:38:33 PM
>>>>>>>> Subject: Re: [Engine-devel] Time to move to
3.3.0-SNAPSHOT?
>>>>>>>>
>>>>>>>> On 01/30/2013 04:31 PM, Alon Bar-Lev wrote:
>>>>>>>>>
>>>>>>>>> Hi,
>>>>>>>>>
>>>>>>>>> You merge this with the rpm version to 3.3.0 which
is
>>>>>>>>> totally
>>>>>>>>> invalid.
>>>>>>>>> 3.3.0 is a *RELEASE*.
>>>>>>>>> I really don't care what maven approach is, but
please do
>>>>>>>>> not
>>>>>>>>> create issues with product release cycle.
>>>>>>>>>
>>>>>>>>> Alon
>>>>>>>>
>>>>>>>> That is not true. After this patch the default RPM
version
>>>>>>>> is
>>>>>>>> 3.3.0-0.1.$(date +%Y%m%d%H%M%S).
>>>>>>>
>>>>>>> 3.3.0 is a release.
>>>>>>>
>>>>>>> 0.1.* is the BUILD.
>>>>>>
>>>>>> In RPM terms 3.3.0 is the "version", and
"0.1.*" is the
>>>>>> "release",
>>>>>> see here:
>>>>>>
>>>>>> [1]
http://www.rpm.org/max-rpm/s1-rpm-inside-tags.html
>>>>>>
>>>>>>> And make dist produces ovirt-engine-3.3.0.tar.gz which is a
>>>>>>> release.
>>>>>>
>>>>>> ovirt-engine-3.3.0.tar.gz is not a release, it is just the
>>>>>> name
>>>>>> of
>>>>>> a
>>>>>> tar
>>>>>> file generated by the makefile. The release is the act of the
>>>>>> release
>>>>>> managers, on behalf of the community, announcing and
>>>>>> publishing
>>>>>> some
>>>>>> artifacts, maybe including a tarball generated with the
>>>>>> makefile,
>>>>>> maybe
>>>>>> one generated manually, as they see fit.
>>>>>
>>>>> This where you are wrong, the name of the source tarball is
>>>>> very
>>>>> important in open source, as this is what open source project
>>>>> actually releases.
>>>>
>>>> The name of the tarball that is published as part of the release
>>>> is
>>>> important, that I agree, maybe not "very" important, but
>>>> important
>>>> anyhow. But what you are saying is that if the name of the file
>>>> generated by default by the Makefile is not exactly the same as
>>>> the
>>>> tarball that will be published as part of the release then it is
>>>> "totally invalid". That is going too far, in my opinion.
>>>
>>> It is important as the version not only appear at the tarball but
>>> within resources such as script --version or Help->About.
>>>
>>> It is also important as people generate their own builds from
>>> tags,
>>> and expect to get proper version of tarball and content within.
>>>
>>> This is not new, this is common to most of the open source world,
>>> oVirt is the exception in this case.
>>
>> All of those are nice goals. None of them has been made impossible
>> by
>> this patch, only that we are using a cleaner maven versioning
>> schema,
>> which is also a nice goal.
>>
>
> The maven version is not exposed to the outside world, for all I
> know it can be 0.0.0 for all past and future versions.
The maven version is exposed to the outside world, not much today,
only
that the POM files are published in every installation done in
Fedora.
And it will be much more exposed in the future, when other people
starts
to develop components based in our (yet to be created) plugin
infrastructure. So it can't be 0.0.0 for all versions.
When we make public API, we stabilize an interface and publish versioned artifacts.
Please notice that the version will not be the product version but the interface version.
For example engine-api-1.0.0 will be available at ovirt-engine-3.2.0, ovirt-engine-3.3.0
and ovirt-engine-4.0.0, as what the developer cares about is the interface version and not
the product version.
We do not have any consumer right now, so maven version is not important, nor the snapshot
attribute used now.
So as far as I understand the maven version is "also a nice goal" but does not
has any actual effect.
> Creating a relation between the maven version and the packaging
> version and not vise versa was what I discussed before you applied
> this patch, and you are CC on the discussions regarding that.
This patch didn't create any relation between maven and packaging
version, it was already there, and I think is reasonable to have it
(to
the extent that packaging systems allow). So if a decision is made to
release 3.3.0-beta2, for example, the corresponding branch should
have a
patch to change the maven version to 3.3.0-beta2, and the packaging
version to 3.3.0-beta2 as well.
I will create such a patch, after we complete the discussion.
> I would have preferred we set proper versioning scheme for 3.3.0
> for the entire package.
> Of course we can fix everything, remove the published snapshots,
> and start over, I would have liked to avoid that.
>
What do you mean by "published snapshots"? As far as I know we don't
publish maven artifacts to any public repository at the moment. Or do
you mean nightly builds?
Yes.
Again, I don't care about maven, as long as not use by any other component it is
meaningless.
>>>
>>> Why do we want to be an exception?
>>>
>>>>
>>>>>
>>>>> We will discuss this offline.
>>>>>
>>>>>>
>>>>>>> You are confusing between upstream and downstream, we will
>>>>>>> have
>>>>>>> to
>>>>>>> fix this before we support more distributions.
>>>>>>
>>>>>> I may be confused, of course, but I don't see the
connection
>>>>>> between
>>>>>> the
>>>>>> name of the tarball generated by the makefile and supporting
>>>>>> more
>>>>>> distributions, looks to me like the least important thing in
>>>>>> these
>>>>>> regards.
>>>>>>
>>>>>>>
>>>>>>>>
>>>>>>>>>
>>>>>>>>> ----- Original Message -----
>>>>>>>>>> From: "Juan Hernandez"
<jhernand(a)redhat.com>
>>>>>>>>>> Cc: engine-devel(a)ovirt.org
>>>>>>>>>> Sent: Wednesday, January 30, 2013 5:26:25 PM
>>>>>>>>>> Subject: Re: [Engine-devel] Time to move to
>>>>>>>>>> 3.3.0-SNAPSHOT?
>>>>>>>>>>
>>>>>>>>>> Note that this change has just been merged. Let
me know if
>>>>>>>>>> you
>>>>>>>>>> find
>>>>>>>>>> any
>>>>>>>>>> issue.
--
Dirección Comercial: C/Jose Bardasano Baos, 9, Edif. Gorbea 3, planta
3ºD, 28016 Madrid, Spain
Inscrita en el Reg. Mercantil de Madrid – C.I.F. B82657941 - Red Hat
S.L.