EA versioning

[Sergi]

Anton,

Since we are not going to mix *b1 *with *beta1* this should work well.

Nevertheless It seems that it will work well for OSGi only if we will mark
our releases as *final*, but I believe
it will be enough to mark them this way only in OSGI Bundle-Version, I
think someone involved in Ignite OSGi
efforts can clarify this better.

http://versionatorr.appspot.com/?a=1.5.0-b1&b=1.5.0-final
http://versionatorr.appspot.com/?a=1.5.0-b1&b=1.5.0

Sergi


2015-12-01 18:40 GMT+03:00 Anton Vinogradov <[hidden email]>:

> Sergi,
>
> Is *b1* equals to *beta1* in OSGi case?
>
> On Tue, Dec 1, 2015 at 6:38 PM, Dmitriy Setrakyan <[hidden email]>
> wrote:
>
> > I like “1.5.0-b1”, etc. Let’s use this for now, as it seems to work.
> >
> > On Tue, Dec 1, 2015 at 7:27 AM, Sergi Vladykin <[hidden email]
> >
> > wrote:
> >
> > > The most promising way which should work for both Maven and OSGi is to
> > use
> > > *beta* for EA and use *final *for releases.
> > >
> > > http://versionatorr.appspot.com/?a=1.5.0-final&b=1.5.0-beta1
> > > http://versionatorr.appspot.com/?a=1.5.0&b=1.5.0-beta1
> > >
> > > Sergi
> > >
> > >
> > > 2015-12-01 18:16 GMT+03:00 Sergi Vladykin <[hidden email]>:
> > >
> > > > Dmitriy,
> > > >
> > > > Yakov just sent maven output which says that
> > > >
> > > > *1.5.0 < 1.5.0-EA1*
> > > >
> > > > *1.5.0-EA1 > 1.5.0-final*
> > > >
> > > > That's exactly what I was talking about.
> > > >
> > > > Sergi
> > > >
> > > > 2015-12-01 18:12 GMT+03:00 Dmitriy Setrakyan <[hidden email]
> >:
> > > >
> > > >> AFAIK, in maven you can have versions with qualifiers, like
> 1.5.0-ea1
> > > and
> > > >> standard versions, like 1.5.0.
> > > >>
> > > >> According to this article [1], all the versions with a qualifier,
> such
> > > as
> > > >> “-ea” will be considered older than the versions without qualifiers.
> > > This
> > > >> means that 1.5.0-ea1 will be older than 1.5.0. Seems like the
> > versioning
> > > >> scheme proposed by Yakov would work.
> > > >>
> > > >> [1] -
> > > >>
> > > >>
> > >
> >
> https://docs.oracle.com/middleware/1212/core/MAVEN/maven_version.htm#MAVEN8855
> > > >>
> > > >> D.
> > > >>
> > > >> On Tue, Dec 1, 2015 at 7:10 AM, Yakov Zhdanov <[hidden email]>
> > > >> wrote:
> > > >>
> > > >> > I have got the following output:
> > > >> >
> > > >> > $ java -cp maven-core-3.3.9.jar:maven-artifact-3.3.9.jar
> > > >> > org.apache.maven.artifact.versioning.ComparableVersion 1.5.0
> > 1.5.0-EA1
> > > >> > 1.5.0-final
> > > >> >
> > > >> > Display parameters as parsed by Maven (in canonical form) and
> > > comparison
> > > >> > result:
> > > >> >
> > > >> > 1. 1.5.0 == 1.5
> > > >> >
> > > >> >    1.5.0 < 1.5.0-EA1
> > > >> >
> > > >> > 2. 1.5.0-EA1 == 1.5-ea-1
> > > >> >
> > > >> >    1.5.0-EA1 > 1.5.0-final
> > > >> >
> > > >> > 3. 1.5.0-final == 1.5
> > > >> >
> > > >> > --Yakov
> > > >> >
> > > >> > 2015-12-01 18:07 GMT+03:00 Sergi Vladykin <
> [hidden email]
> > >:
> > > >> >
> > > >> > > Also it is an interesting subject with respect to OSGi
> versioning
> > > >> because
> > > >> > > Maven and OSGi versions are
> > > >> > > somewhat conflicting as well. See [1]
> > > >> > >
> > > >> > > [1] http://versionatorr.appspot.com/?a=1.5.0-final&b=1.5.0-ea
> > > >> > >
> > > >> > > Sergi
> > > >> > >
> > > >> > > 2015-12-01 17:56 GMT+03:00 Raul Kripalani <[hidden email]>:
> > > >> > >
> > > >> > > > It is different when the keyword is part of the version
> > > >> > (1.1.1.RELEASE),
> > > >> > > > like Spring, and when it's a qualifier (1.1.1-RELEASE).
> > > >> > > >
> > > >> > > > Maven treats both cases differently.
> > > >> > > > On 1 Dec 2015 14:52, "Yakov Zhdanov" <[hidden email]>
> > wrote:
> > > >> > > >
> > > >> > > > > Sergi, very good point! Guys, it seems that EA is not a good
> > > >> choice.
> > > >> > > > >
> > > >> > > > > However, how many of you have ever used RELEASE as version
> in
> > > >> maven
> > > >> > or
> > > >> > > > > version range?
> > > >> > > > >
> > > >> > > > > --Yakov
> > > >> > > > >
> > > >> > > >
> > > >> > >
> > > >> >
> > > >>
> > > >
> > > >
> > >
> >
>

[Sergi]

FYI, OSGi sorts version qualifiers lexicographically:

http://wiki.osgi.org/wiki/Bundle-Version

Sergi

2015-12-01 18:55 GMT+03:00 Sergi Vladykin <[hidden email]>:

> Anton,
>
> Since we are not going to mix *b1 *with *beta1* this should work well.
>
> Nevertheless It seems that it will work well for OSGi only if we will mark
> our releases as *final*, but I believe
> it will be enough to mark them this way only in OSGI Bundle-Version, I
> think someone involved in Ignite OSGi
> efforts can clarify this better.
>
> http://versionatorr.appspot.com/?a=1.5.0-b1&b=1.5.0-final
> http://versionatorr.appspot.com/?a=1.5.0-b1&b=1.5.0
>
> Sergi
>
>
>
> 2015-12-01 18:40 GMT+03:00 Anton Vinogradov <[hidden email]>:
>
>> Sergi,
>>
>> Is *b1* equals to *beta1* in OSGi case?
>>
>> On Tue, Dec 1, 2015 at 6:38 PM, Dmitriy Setrakyan <[hidden email]>
>> wrote:
>>
>> > I like “1.5.0-b1”, etc. Let’s use this for now, as it seems to work.
>> >
>> > On Tue, Dec 1, 2015 at 7:27 AM, Sergi Vladykin <
>> [hidden email]>
>> > wrote:
>> >
>> > > The most promising way which should work for both Maven and OSGi is to
>> > use
>> > > *beta* for EA and use *final *for releases.
>> > >
>> > > http://versionatorr.appspot.com/?a=1.5.0-final&b=1.5.0-beta1
>> > > http://versionatorr.appspot.com/?a=1.5.0&b=1.5.0-beta1
>> > >
>> > > Sergi
>> > >
>> > >
>> > > 2015-12-01 18:16 GMT+03:00 Sergi Vladykin <[hidden email]>:
>> > >
>> > > > Dmitriy,
>> > > >
>> > > > Yakov just sent maven output which says that
>> > > >
>> > > > *1.5.0 < 1.5.0-EA1*
>> > > >
>> > > > *1.5.0-EA1 > 1.5.0-final*
>> > > >
>> > > > That's exactly what I was talking about.
>> > > >
>> > > > Sergi
>> > > >
>> > > > 2015-12-01 18:12 GMT+03:00 Dmitriy Setrakyan <[hidden email]
>> >:
>> > > >
>> > > >> AFAIK, in maven you can have versions with qualifiers, like
>> 1.5.0-ea1
>> > > and
>> > > >> standard versions, like 1.5.0.
>> > > >>
>> > > >> According to this article [1], all the versions with a qualifier,
>> such
>> > > as
>> > > >> “-ea” will be considered older than the versions without
>> qualifiers.
>> > > This
>> > > >> means that 1.5.0-ea1 will be older than 1.5.0. Seems like the
>> > versioning
>> > > >> scheme proposed by Yakov would work.
>> > > >>
>> > > >> [1] -
>> > > >>
>> > > >>
>> > >
>> >
>> https://docs.oracle.com/middleware/1212/core/MAVEN/maven_version.htm#MAVEN8855
>> > > >>
>> > > >> D.
>> > > >>
>> > > >> On Tue, Dec 1, 2015 at 7:10 AM, Yakov Zhdanov <[hidden email]
>> >
>> > > >> wrote:
>> > > >>
>> > > >> > I have got the following output:
>> > > >> >
>> > > >> > $ java -cp maven-core-3.3.9.jar:maven-artifact-3.3.9.jar
>> > > >> > org.apache.maven.artifact.versioning.ComparableVersion 1.5.0
>> > 1.5.0-EA1
>> > > >> > 1.5.0-final
>> > > >> >
>> > > >> > Display parameters as parsed by Maven (in canonical form) and
>> > > comparison
>> > > >> > result:
>> > > >> >
>> > > >> > 1. 1.5.0 == 1.5
>> > > >> >
>> > > >> >    1.5.0 < 1.5.0-EA1
>> > > >> >
>> > > >> > 2. 1.5.0-EA1 == 1.5-ea-1
>> > > >> >
>> > > >> >    1.5.0-EA1 > 1.5.0-final
>> > > >> >
>> > > >> > 3. 1.5.0-final == 1.5
>> > > >> >
>> > > >> > --Yakov
>> > > >> >
>> > > >> > 2015-12-01 18:07 GMT+03:00 Sergi Vladykin <
>> [hidden email]
>> > >:
>> > > >> >
>> > > >> > > Also it is an interesting subject with respect to OSGi
>> versioning
>> > > >> because
>> > > >> > > Maven and OSGi versions are
>> > > >> > > somewhat conflicting as well. See [1]
>> > > >> > >
>> > > >> > > [1] http://versionatorr.appspot.com/?a=1.5.0-final&b=1.5.0-ea
>> > > >> > >
>> > > >> > > Sergi
>> > > >> > >
>> > > >> > > 2015-12-01 17:56 GMT+03:00 Raul Kripalani <[hidden email]>:
>> > > >> > >
>> > > >> > > > It is different when the keyword is part of the version
>> > > >> > (1.1.1.RELEASE),
>> > > >> > > > like Spring, and when it's a qualifier (1.1.1-RELEASE).
>> > > >> > > >
>> > > >> > > > Maven treats both cases differently.
>> > > >> > > > On 1 Dec 2015 14:52, "Yakov Zhdanov" <[hidden email]>
>> > wrote:
>> > > >> > > >
>> > > >> > > > > Sergi, very good point! Guys, it seems that EA is not a
>> good
>> > > >> choice.
>> > > >> > > > >
>> > > >> > > > > However, how many of you have ever used RELEASE as version
>> in
>> > > >> maven
>> > > >> > or
>> > > >> > > > > version range?
>> > > >> > > > >
>> > > >> > > > > --Yakov
>> > > >> > > > >
>> > > >> > > >
>> > > >> > >
>> > > >> >
>> > > >>
>> > > >
>> > > >
>> > >
>> >
>>
>
>

[Raul Kripalani]

On Tue, Dec 1, 2015 at 3:33 PM, Yakov Zhdanov <[hidden email]> wrote:

> 1.5.0-ea > 1.5.0


Am I the only one to whom this sounds wrong?

*Raúl Kripalani*
PMC & Committer @ Apache Ignite, Apache Camel | Integration, Big Data and
Messaging Engineer
http://about.me/raulkripalani | http://www.linkedin.com/in/raulkripalani
http://blog.raulkr.net | twitter: @raulvk

[Sergi]

Raul,

This is how Maven works, exactly because of that we want to switch to
1.5.0-b1 instead of 1.5.0-ea

1.5.0 > 1.5.0-b1

Sergi

2015-12-01 19:06 GMT+03:00 Raul Kripalani <[hidden email]>:

> On Tue, Dec 1, 2015 at 3:33 PM, Yakov Zhdanov <[hidden email]> wrote:
>
> > 1.5.0-ea > 1.5.0
>
>
> Am I the only one to whom this sounds wrong?
>
> *Raúl Kripalani*
> PMC & Committer @ Apache Ignite, Apache Camel | Integration, Big Data and
> Messaging Engineer
> http://about.me/raulkripalani | http://www.linkedin.com/in/raulkripalani
> http://blog.raulkr.net | twitter: @raulvk
>

[Raul Kripalani]

Let me synthesise:

* 1.5.0-ea, 1.5.0 => comparison result is off in Maven, because "ea" is not
a well-known qualifier [0] to Maven, therefore it is thought higher than no
qualifier. Same goes with 1.5.0-ea, 1.5.0-final.
* 1.5.0-b1, 1.5.0 => mismatch between Maven and OSGi [1], because while
Maven treats b1 as a known qualifier, OSGi treats it lexicographically.
* 1.5.0-b1, 1.5.0-final => would work for both Maven and OSGi [2].

I'm in favour of the following scheme: 1.5.0-a1, 1.5.0-b2, 1.5.0-final,
1.5.0-sp1.

Explicitly prohibiting the following because they are previous to "final",
but they'd be considered newer by OSGi due to lexicographic evaluation:

 * 1.5.0-rc1
 * 1.5.0-m1

We do have an edge case with snapshot, which is older than final, but it'd
always be considered newer by OSGi. I don't think it's a big deal, though.

With this (now well-) defined scheme, I would be happy to publish alphas,
betas and service packs to Maven Central.

If you'd like to continue chatting, I'm in Gitter:
https://gitter.im/apacheignite/ignite.

[0]
https://github.com/apache/maven/blob/master/maven-artifact/src/main/java/org/apache/maven/artifact/versioning/ComparableVersion.java
[1] https://versionatorr.appspot.com/?a=1.5.0&b=1.5.0-b1
[2] https://versionatorr.appspot.com/?a=1.5.0-b1&b=1.5.0-final

*Raúl Kripalani*
PMC & Committer @ Apache Ignite, Apache Camel | Integration, Big Data and
Messaging Engineer
http://about.me/raulkripalani | http://www.linkedin.com/in/raulkripalani
http://blog.raulkr.net | twitter: @raulvk

On Tue, Dec 1, 2015 at 4:13 PM, Sergi Vladykin <[hidden email]>
wrote:

> Raul,
>
> This is how Maven works, exactly because of that we want to switch to
> 1.5.0-b1 instead of 1.5.0-ea
>
> 1.5.0 > 1.5.0-b1
>
> Sergi
>
> 2015-12-01 19:06 GMT+03:00 Raul Kripalani <[hidden email]>:
>
> > On Tue, Dec 1, 2015 at 3:33 PM, Yakov Zhdanov <[hidden email]>
> wrote:
> >
> > > 1.5.0-ea > 1.5.0
> >
> >
> > Am I the only one to whom this sounds wrong?
> >
> > *Raúl Kripalani*
> > PMC & Committer @ Apache Ignite, Apache Camel | Integration, Big Data and
> > Messaging Engineer
> > http://about.me/raulkripalani | http://www.linkedin.com/in/raulkripalani
> > http://blog.raulkr.net | twitter: @raulvk
> >
>

[Sergi]

Thanks, Raul! I like your synthesis.

We have already agreed to use 1.5.0-*b1* format for early access builds,
but I did not see any opinions about *final* for releases.
Does anyone have objections to use *final *qualifier for release builds to
conform both OSGi and Maven versioning?

Sergi

2015-12-01 19:45 GMT+03:00 Raul Kripalani <[hidden email]>:

> Let me synthesise:
>
> * 1.5.0-ea, 1.5.0 => comparison result is off in Maven, because "ea" is not
> a well-known qualifier [0] to Maven, therefore it is thought higher than no
> qualifier. Same goes with 1.5.0-ea, 1.5.0-final.
> * 1.5.0-b1, 1.5.0 => mismatch between Maven and OSGi [1], because while
> Maven treats b1 as a known qualifier, OSGi treats it lexicographically.
> * 1.5.0-b1, 1.5.0-final => would work for both Maven and OSGi [2].
>
> I'm in favour of the following scheme: 1.5.0-a1, 1.5.0-b2, 1.5.0-final,
> 1.5.0-sp1.
>
> Explicitly prohibiting the following because they are previous to "final",
> but they'd be considered newer by OSGi due to lexicographic evaluation:
>
>  * 1.5.0-rc1
>  * 1.5.0-m1
>
> We do have an edge case with snapshot, which is older than final, but it'd
> always be considered newer by OSGi. I don't think it's a big deal, though.
>
> With this (now well-) defined scheme, I would be happy to publish alphas,
> betas and service packs to Maven Central.
>
> If you'd like to continue chatting, I'm in Gitter:
> https://gitter.im/apacheignite/ignite.
>
> [0]
>
> https://github.com/apache/maven/blob/master/maven-artifact/src/main/java/org/apache/maven/artifact/versioning/ComparableVersion.java
> [1] https://versionatorr.appspot.com/?a=1.5.0&b=1.5.0-b1
> [2] https://versionatorr.appspot.com/?a=1.5.0-b1&b=1.5.0-final
>
> *Raúl Kripalani*
> PMC & Committer @ Apache Ignite, Apache Camel | Integration, Big Data and
> Messaging Engineer
> http://about.me/raulkripalani | http://www.linkedin.com/in/raulkripalani
> http://blog.raulkr.net | twitter: @raulvk
>
> On Tue, Dec 1, 2015 at 4:13 PM, Sergi Vladykin <[hidden email]>
> wrote:
>
> > Raul,
> >
> > This is how Maven works, exactly because of that we want to switch to
> > 1.5.0-b1 instead of 1.5.0-ea
> >
> > 1.5.0 > 1.5.0-b1
> >
> > Sergi
> >
> > 2015-12-01 19:06 GMT+03:00 Raul Kripalani <[hidden email]>:
> >
> > > On Tue, Dec 1, 2015 at 3:33 PM, Yakov Zhdanov <[hidden email]>
> > wrote:
> > >
> > > > 1.5.0-ea > 1.5.0
> > >
> > >
> > > Am I the only one to whom this sounds wrong?
> > >
> > > *Raúl Kripalani*
> > > PMC & Committer @ Apache Ignite, Apache Camel | Integration, Big Data
> and
> > > Messaging Engineer
> > > http://about.me/raulkripalani |
> http://www.linkedin.com/in/raulkripalani
> > > http://blog.raulkr.net | twitter: @raulvk
> > >
> >
>

[Raul Kripalani]

On Tue, Dec 1, 2015 at 5:57 PM, Sergi Vladykin <[hidden email]>
wrote:

> Thanks, Raul! I like your synthesis.
>
> We have already agreed to use 1.5.0-*b1* format for early access builds,
> but I did not see any opinions about *final* for releases.
> Does anyone have objections to use *final *qualifier for release builds to
> conform both OSGi and Maven versioning?
>

There is also the possibility to drop the qualifier in package exports
(Bundle-Version can stay as-is).
But I don't like this, we'd be losing information, it leads to possibly
incoherent environments and it's confusing.
I just wanted to mention it for the sake of completeness.

*Raúl Kripalani*
PMC & Committer @ Apache Ignite, Apache Camel | Integration, Big Data and
Messaging Engineer
http://about.me/raulkripalani | http://www.linkedin.com/in/raulkripalani
http://blog.raulkr.net | twitter: @raulvk

[dsetrakyan]

What would the “final” qualifier give us? From what I can see, Maven
version comparison will work fine without it. Does maven have special
handling for the “final” suffix?

On Tue, Dec 1, 2015 at 10:36 AM, Raul Kripalani <[hidden email]> wrote:

> On Tue, Dec 1, 2015 at 5:57 PM, Sergi Vladykin <[hidden email]>
> wrote:
>
> > Thanks, Raul! I like your synthesis.
> >
> > We have already agreed to use 1.5.0-*b1* format for early access builds,
> > but I did not see any opinions about *final* for releases.
> > Does anyone have objections to use *final *qualifier for release builds
> to
> > conform both OSGi and Maven versioning?
> >
>
> There is also the possibility to drop the qualifier in package exports
> (Bundle-Version can stay as-is).
> But I don't like this, we'd be losing information, it leads to possibly
> incoherent environments and it's confusing.
> I just wanted to mention it for the sake of completeness.
>
> *Raúl Kripalani*
> PMC & Committer @ Apache Ignite, Apache Camel | Integration, Big Data and
> Messaging Engineer
> http://about.me/raulkripalani | http://www.linkedin.com/in/raulkripalani
> http://blog.raulkr.net | twitter: @raulvk
>

[Raul Kripalani]

On Tue, Dec 1, 2015 at 6:55 PM, Dmitriy Setrakyan <[hidden email]>
wrote:

> What would the “final” qualifier give us? From what I can see, Maven
> version comparison will work fine without it. Does maven have special
> handling for the “final” suffix?
>

Yes it does. And OSGi doesn't, so a version range like [1.5.0.b1,2) would
not include 1.5.0.

Regards,

*Raúl Kripalani*
PMC & Committer @ Apache Ignite, Apache Camel | Integration, Big Data and
Messaging Engineer
http://about.me/raulkripalani | http://www.linkedin.com/in/raulkripalani
http://blog.raulkr.net | twitter: @raulvk

[dsetrakyan]

On Tue, Dec 1, 2015 at 11:09 AM, Raul Kripalani <[hidden email]> wrote:

> On Tue, Dec 1, 2015 at 6:55 PM, Dmitriy Setrakyan <[hidden email]>
> wrote:
>
> > What would the “final” qualifier give us? From what I can see, Maven
> > version comparison will work fine without it. Does maven have special
> > handling for the “final” suffix?
>

In that case, if we add “.final” suffix, will it satisfy both, Maven and
OSGI worlds?

> >
>
> Yes it does. And OSGi doesn't, so a version range like [1.5.0.b1,2) would
> not include 1.5.0.
>
> Regards,
>
> *Raúl Kripalani*
> PMC & Committer @ Apache Ignite, Apache Camel | Integration, Big Data and
> Messaging Engineer
> http://about.me/raulkripalani | http://www.linkedin.com/in/raulkripalani
> http://blog.raulkr.net | twitter: @raulvk
>

[Sergi]

2015-12-01 21:55 GMT+03:00 Dmitriy Setrakyan <[hidden email]>:

> What would the “final” qualifier give us?


It gives us correct and consistent versions not only for Maven, but for
OSGi as well.

Now we will have
For Maven      1.5.0 > 1.5.0-b1
For OSGi        1.5.0 < 1.5.0-b1

If we will add *final* qualifier we will have
For Maven      1.5.0-final > 1.5.0-b1
For OSGi        1.5.0-final > 1.5.0-b1


> From what I can see, Maven
> version comparison will work fine without it. Does maven have special
> handling for the “final” suffix?
>
>
Yes, it is the same as having no suffix at all, e.g. 1.5.0 == 1.5.0-final

Sergi

> On Tue, Dec 1, 2015 at 10:36 AM, Raul Kripalani <[hidden email]> wrote:
>
> > On Tue, Dec 1, 2015 at 5:57 PM, Sergi Vladykin <[hidden email]
> >
> > wrote:
> >
> > > Thanks, Raul! I like your synthesis.
> > >
> > > We have already agreed to use 1.5.0-*b1* format for early access
> builds,
> > > but I did not see any opinions about *final* for releases.
> > > Does anyone have objections to use *final *qualifier for release builds
> > to
> > > conform both OSGi and Maven versioning?
> > >
> >
> > There is also the possibility to drop the qualifier in package exports
> > (Bundle-Version can stay as-is).
> > But I don't like this, we'd be losing information, it leads to possibly
> > incoherent environments and it's confusing.
> > I just wanted to mention it for the sake of completeness.
> >
> > *Raúl Kripalani*
> > PMC & Committer @ Apache Ignite, Apache Camel | Integration, Big Data and
> > Messaging Engineer
> > http://about.me/raulkripalani | http://www.linkedin.com/in/raulkripalani
> > http://blog.raulkr.net | twitter: @raulvk
> >
>

[dsetrakyan]

Thanks Sergi,

In that case I don’t think anyone would object to adding “.final” suffix at
the end.

Raul, as an OSGI expert, do you confirm?

D.

On Tue, Dec 1, 2015 at 11:13 AM, Sergi Vladykin <[hidden email]>
wrote:

> 2015-12-01 21:55 GMT+03:00 Dmitriy Setrakyan <[hidden email]>:
>
> > What would the “final” qualifier give us?
>
>
> It gives us correct and consistent versions not only for Maven, but for
> OSGi as well.
>
> Now we will have
> For Maven      1.5.0 > 1.5.0-b1
> For OSGi        1.5.0 < 1.5.0-b1
>
> If we will add *final* qualifier we will have
> For Maven      1.5.0-final > 1.5.0-b1
> For OSGi        1.5.0-final > 1.5.0-b1
>
>
> > From what I can see, Maven
> > version comparison will work fine without it. Does maven have special
> > handling for the “final” suffix?
> >
> >
> Yes, it is the same as having no suffix at all, e.g. 1.5.0 == 1.5.0-final
>
> Sergi
>
>
> > On Tue, Dec 1, 2015 at 10:36 AM, Raul Kripalani <[hidden email]>
> wrote:
> >
> > > On Tue, Dec 1, 2015 at 5:57 PM, Sergi Vladykin <
> [hidden email]
> > >
> > > wrote:
> > >
> > > > Thanks, Raul! I like your synthesis.
> > > >
> > > > We have already agreed to use 1.5.0-*b1* format for early access
> > builds,
> > > > but I did not see any opinions about *final* for releases.
> > > > Does anyone have objections to use *final *qualifier for release
> builds
> > > to
> > > > conform both OSGi and Maven versioning?
> > > >
> > >
> > > There is also the possibility to drop the qualifier in package exports
> > > (Bundle-Version can stay as-is).
> > > But I don't like this, we'd be losing information, it leads to possibly
> > > incoherent environments and it's confusing.
> > > I just wanted to mention it for the sake of completeness.
> > >
> > > *Raúl Kripalani*
> > > PMC & Committer @ Apache Ignite, Apache Camel | Integration, Big Data
> and
> > > Messaging Engineer
> > > http://about.me/raulkripalani |
> http://www.linkedin.com/in/raulkripalani
> > > http://blog.raulkr.net | twitter: @raulvk
> > >
> >
>

[Raul Kripalani]

On Tue, Dec 1, 2015 at 7:12 PM, Dmitriy Setrakyan <[hidden email]>
wrote:

>
> In that case, if we add “.final” suffix, will it satisfy both, Maven and
> OSGI worlds?


Yep. In fact, "final" in Maven is equivalent to no qualifier. From Javadoc
[1]:

  strings are checked for well-known qualifiers and the qualifier ordering
is used for version ordering. Well-known qualifiers (case insensitive) are:
     [...]
     (the empty string) or ga or final
     [...]

So it should have no side effects, other than the possible (very minor)
nuisance for some having to suffix version references with -final.

[1]
https://maven.apache.org/ref/3.3.1/maven-artifact/apidocs/index.html?org/apache/maven/artifact/versioning/ComparableVersion.html

*Raúl Kripalani*
PMC & Committer @ Apache Ignite, Apache Camel | Integration, Big Data and
Messaging Engineer
http://about.me/raulkripalani | http://www.linkedin.com/in/raulkripalani
http://blog.raulkr.net | twitter: @raulvk

[Raul Kripalani]

On Tue, Dec 1, 2015 at 7:15 PM, Dmitriy Setrakyan <[hidden email]>
wrote:

> Raul, as an OSGI expert, do you confirm?
>

Yep, it was my proposal only to add "-final". Just to be clear, this is a
Maven qualifier. The maven-bundle-plugin will translate the hyphen to a
dot, for compatibility with OSGi.

*Raúl Kripalani*
PMC & Committer @ Apache Ignite, Apache Camel | Integration, Big Data and
Messaging Engineer
http://about.me/raulkripalani | http://www.linkedin.com/in/raulkripalani
http://blog.raulkr.net | twitter: @raulvk

[yzhdanov]

Guys, I like the scheme we have come to.

Will submit apache-ignite-1.5.0-b1-rc1 for vote tomorrow :)

--Yakov

[Raul Kripalani]

Ok from me, the troublemaker :)

Glad that we agreed on a policy.
On 1 Dec 2015 19:51, "Yakov Zhdanov" <[hidden email]> wrote:

> Guys, I like the scheme we have come to.
>
> Will submit apache-ignite-1.5.0-b1-rc1 for vote tomorrow :)
>
> --Yakov
>

[yzhdanov]

Thanks you, Raul, for attracting everyones attention to! :)

--Yakov

2015-12-02 0:15 GMT+03:00 Raul Kripalani <[hidden email]>:

> Ok from me, the troublemaker :)
>
> Glad that we agreed on a policy.

[Alexey Kuznetsov-2]

Guys,

We are introduced new version naming strategy, but do not added
corresponding tests to GridProductVersionSelfTest.

I created issue for this:https://issues.apache.org/jira/browse/IGNITE-2049
Please fix before ignite-1.5 goes final.

--
Alexey Kuznetsov
GridGain Systems
www.gridgain.com

[Romain Gilles-2]

Hi Igniter,
Just to be a pain ...
I like the Spring (I think derive from Eclipse) version numbering approach.
And it works like a charm with OSGi and maven.

For the sprint they deliver milestone : 1.6.0.M1, 1.6.0.M2...
MX imply no guaranty of stable model / classes.

When they enter into release process with release branch they use:
1.6.0.RC1 ...
RC imply stable model only bug fix

And finally when the big day comes they use the RELEASE qualifier =>
1.6.0.RELEASE.

Not sure it help but I just like it and want to let you know.

Romain

Le mer. 2 déc. 2015 à 04:37, Alexey Kuznetsov <[hidden email]> a
écrit :

> Guys,
>
> We are introduced new version naming strategy, but do not added
> corresponding tests to GridProductVersionSelfTest.
>
> I created issue for this:https://issues.apache.org/jira/browse/IGNITE-2049
> Please fix before ignite-1.5 goes final.
>
> --
> Alexey Kuznetsov
> GridGain Systems
> www.gridgain.com
>

[Sergi]

Romain,

I believe it is a good idea to stick to well-known Maven qualifiers which
already exist.
*RELEASE* is not in this list and I'd prefer to avoid surprises, if this
will be changed later.
Also our release cycle is different from Eclipse and Spring (instead of
milestones
we have betas, but RC has absolutely another meaning for us).
Thus I don't see any reasons to reflect their versioning in our product.

Sergi

2015-12-03 18:21 GMT+03:00 Romain Gilles <[hidden email]>:

> Hi Igniter,
> Just to be a pain ...
> I like the Spring (I think derive from Eclipse) version numbering approach.
> And it works like a charm with OSGi and maven.
>
> For the sprint they deliver milestone : 1.6.0.M1, 1.6.0.M2...
> MX imply no guaranty of stable model / classes.
>
> When they enter into release process with release branch they use:
> 1.6.0.RC1 ...
> RC imply stable model only bug fix
>
> And finally when the big day comes they use the RELEASE qualifier =>
> 1.6.0.RELEASE.
>
> Not sure it help but I just like it and want to let you know.
>
> Romain
>
> Le mer. 2 déc. 2015 à 04:37, Alexey Kuznetsov <[hidden email]> a
> écrit :
>
> > Guys,
> >
> > We are introduced new version naming strategy, but do not added
> > corresponding tests to GridProductVersionSelfTest.
> >
> > I created issue for this:
> https://issues.apache.org/jira/browse/IGNITE-2049
> > Please fix before ignite-1.5 goes final.
> >
> > --
> > Alexey Kuznetsov
> > GridGain Systems
> > www.gridgain.com
> >
>