[math] Version mgt idea

classic Classic list List threaded Threaded
33 messages Options
12
Reply | Threaded
Open this post in threaded view
|

[math] Version mgt idea

Phil Steitz
Here is an idea that might break our deadlock re backward
compatibility, versioning and RERO:

Agree that odd numbered versions have stable APIs - basically adhere
to Commons rules - no breaks within 3.0, 3.1, ..., 3.x... or 5.0,
5.1... but even-numbered lines can include breaks - so 4.0 and 4.1
might not be compatible.  We would always maintain both an odd and
even branch - ideally in such a way that when an even numbered line
stabilized it would add a last hurrah of breaks and move to odd.
People wanting stable APIs could just stick with the odd-numbered
lines and [math] developers wanting to experiment with things and
not worry about compatibility could do that in the even-numbered
lines.  In effect, this is sort of what we are doing now in 3.x / 4.x.

I know above violates Commons policy if we actually cut releases
from the even-numbered branches - we would have to get agreement
from the Commons PMC that this is OK or somehow label the releases
differently.  Just an idea to get us out of our current bind...

Phil


---------------------------------------------------------------------
To unsubscribe, e-mail: [hidden email]
For additional commands, e-mail: [hidden email]

Reply | Threaded
Open this post in threaded view
|

Re: [math] Version mgt idea

sebb-2-2
On 6 November 2015 at 16:17, Phil Steitz <[hidden email]> wrote:

> Here is an idea that might break our deadlock re backward
> compatibility, versioning and RERO:
>
> Agree that odd numbered versions have stable APIs - basically adhere
> to Commons rules - no breaks within 3.0, 3.1, ..., 3.x... or 5.0,
> 5.1... but even-numbered lines can include breaks - so 4.0 and 4.1
> might not be compatible.  We would always maintain both an odd and
> even branch - ideally in such a way that when an even numbered line
> stabilized it would add a last hurrah of breaks and move to odd.
> People wanting stable APIs could just stick with the odd-numbered
> lines and [math] developers wanting to experiment with things and
> not worry about compatibility could do that in the even-numbered
> lines.  In effect, this is sort of what we are doing now in 3.x / 4.x.
>
> I know above violates Commons policy if we actually cut releases
> from the even-numbered branches - we would have to get agreement
> from the Commons PMC that this is OK or somehow label the releases
> differently.  Just an idea to get us out of our current bind...

That would likely cause problems with Maven, which will pick the
latest release, regardless of whether it is even or odd.
[However changing the gid/uid without changing the package name also
causes problems with Maven.]

Unless it is possible to add a marker to the release such that Maven
does not consider it when looking for the latest release.
The only such marker I know of is SNAPSHOT - such releases are not
added to Maven Central so cannot be accidentally added to the
classpath.

Maybe there is another marker which could be used that tells Maven not
to consider that version.

If there is no other such marker, one way to avoid these issues would
be to not publish the even releases to Maven Central.
Users would have to download them some other way.

Maybe there is mileage in having a staging repo for Commons as a half-way house.
Users would need to specifically add the repo to their pom to use the repo.
[I think I suggested something like this for some Commons Maven
plugins that were for developer use only]

Note that publishing even-numbered releases to the ASF mirror service
can also cause problems if the unstable releases are not clearly
marked.

> Phil
>
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: [hidden email]
> For additional commands, e-mail: [hidden email]
>

---------------------------------------------------------------------
To unsubscribe, e-mail: [hidden email]
For additional commands, e-mail: [hidden email]

Reply | Threaded
Open this post in threaded view
|

Re: [math] Version mgt idea

Phil Steitz
On 11/6/15 10:18 AM, sebb wrote:

> On 6 November 2015 at 16:17, Phil Steitz <[hidden email]> wrote:
>> Here is an idea that might break our deadlock re backward
>> compatibility, versioning and RERO:
>>
>> Agree that odd numbered versions have stable APIs - basically adhere
>> to Commons rules - no breaks within 3.0, 3.1, ..., 3.x... or 5.0,
>> 5.1... but even-numbered lines can include breaks - so 4.0 and 4.1
>> might not be compatible.  We would always maintain both an odd and
>> even branch - ideally in such a way that when an even numbered line
>> stabilized it would add a last hurrah of breaks and move to odd.
>> People wanting stable APIs could just stick with the odd-numbered
>> lines and [math] developers wanting to experiment with things and
>> not worry about compatibility could do that in the even-numbered
>> lines.  In effect, this is sort of what we are doing now in 3.x / 4.x.
>>
>> I know above violates Commons policy if we actually cut releases
>> from the even-numbered branches - we would have to get agreement
>> from the Commons PMC that this is OK or somehow label the releases
>> differently.  Just an idea to get us out of our current bind...
> That would likely cause problems with Maven, which will pick the
> latest release, regardless of whether it is even or odd.
> [However changing the gid/uid without changing the package name also
> causes problems with Maven.]

I am curious how many people actually use unversioned dependencies.
I would never do that.  But I get the concern that as soon as you
release an even-numbered release it becomes the latest release and
people might just grab it.

>
> Unless it is possible to add a marker to the release such that Maven
> does not consider it when looking for the latest release.
> The only such marker I know of is SNAPSHOT - such releases are not
> added to Maven Central so cannot be accidentally added to the
> classpath.
>
> Maybe there is another marker which could be used that tells Maven not
> to consider that version.
>
> If there is no other such marker, one way to avoid these issues would
> be to not publish the even releases to Maven Central.
> Users would have to download them some other way.
>
> Maybe there is mileage in having a staging repo for Commons as a half-way house.
> Users would need to specifically add the repo to their pom to use the repo.
> [I think I suggested something like this for some Commons Maven
> plugins that were for developer use only]
>
> Note that publishing even-numbered releases to the ASF mirror service
> can also cause problems if the unstable releases are not clearly
> marked.

Right - there we could mark them.  We could make it clear in release
notes, READMEs, web pages, etc.

Phil

>
>> Phil
>>
>>
>> ---------------------------------------------------------------------
>> To unsubscribe, e-mail: [hidden email]
>> For additional commands, e-mail: [hidden email]
>>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: [hidden email]
> For additional commands, e-mail: [hidden email]
>
>


---------------------------------------------------------------------
To unsubscribe, e-mail: [hidden email]
For additional commands, e-mail: [hidden email]

Reply | Threaded
Open this post in threaded view
|

Re: [math] Version mgt idea

Gilles Sadowski
In reply to this post by Phil Steitz
Hi.

On Fri, 6 Nov 2015 09:17:18 -0700, Phil Steitz wrote:
> Here is an idea that might break our deadlock re backward
> compatibility, versioning and RERO:
>
> Agree that odd numbered versions have stable APIs - basically adhere
> to Commons rules - no breaks within 3.0, 3.1, ..., 3.x... or 5.0,
> 5.1... but even-numbered lines can include breaks -

I like the proposal to be more lax on compatibility breaking than
I ever dreamed of.  ;-)

> so 4.0 and 4.1
> might not be compatible.

Isn't that going to cause JAR hell?

Gilles

> We would always maintain both an odd and
> even branch - ideally in such a way that when an even numbered line
> stabilized it would add a last hurrah of breaks and move to odd.
> People wanting stable APIs could just stick with the odd-numbered
> lines and [math] developers wanting to experiment with things and
> not worry about compatibility could do that in the even-numbered
> lines.  In effect, this is sort of what we are doing now in 3.x /
> 4.x.
>
> I know above violates Commons policy if we actually cut releases
> from the even-numbered branches - we would have to get agreement
> from the Commons PMC that this is OK or somehow label the releases
> differently.  Just an idea to get us out of our current bind...
>
> Phil
>
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: [hidden email]
> For additional commands, e-mail: [hidden email]


---------------------------------------------------------------------
To unsubscribe, e-mail: [hidden email]
For additional commands, e-mail: [hidden email]

Reply | Threaded
Open this post in threaded view
|

Re: [math] Version mgt idea

Phil Steitz
On 11/6/15 10:31 AM, Gilles wrote:

> Hi.
>
> On Fri, 6 Nov 2015 09:17:18 -0700, Phil Steitz wrote:
>> Here is an idea that might break our deadlock re backward
>> compatibility, versioning and RERO:
>>
>> Agree that odd numbered versions have stable APIs - basically adhere
>> to Commons rules - no breaks within 3.0, 3.1, ..., 3.x... or 5.0,
>> 5.1... but even-numbered lines can include breaks -
>
> I like the proposal to be more lax on compatibility breaking than
> I ever dreamed of.  ;-)
>
>> so 4.0 and 4.1
>> might not be compatible.
>
> Isn't that going to cause JAR hell?

Yes, within the even-numbered branches.  But we would repackage and
rename at each version cut as we do now.  So the only "hell" would
be if someone deploys multiple different versions from an
even-numbered branch.  We would expect them to be less widely
deployed, so this would be less of an issue.

Phil

>
> Gilles
>
>> We would always maintain both an odd and
>> even branch - ideally in such a way that when an even numbered line
>> stabilized it would add a last hurrah of breaks and move to odd.
>> People wanting stable APIs could just stick with the odd-numbered
>> lines and [math] developers wanting to experiment with things and
>> not worry about compatibility could do that in the even-numbered
>> lines.  In effect, this is sort of what we are doing now in 3.x /
>> 4.x.
>>
>> I know above violates Commons policy if we actually cut releases
>> from the even-numbered branches - we would have to get agreement
>> from the Commons PMC that this is OK or somehow label the releases
>> differently.  Just an idea to get us out of our current bind...
>>
>> Phil
>>
>>
>> ---------------------------------------------------------------------
>>
>> To unsubscribe, e-mail: [hidden email]
>> For additional commands, e-mail: [hidden email]
>
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: [hidden email]
> For additional commands, e-mail: [hidden email]
>
>


---------------------------------------------------------------------
To unsubscribe, e-mail: [hidden email]
For additional commands, e-mail: [hidden email]

Reply | Threaded
Open this post in threaded view
|

Re: [math] Version mgt idea

Gilles Sadowski
On Fri, 6 Nov 2015 10:36:51 -0700, Phil Steitz wrote:

> On 11/6/15 10:31 AM, Gilles wrote:
>> Hi.
>>
>> On Fri, 6 Nov 2015 09:17:18 -0700, Phil Steitz wrote:
>>> Here is an idea that might break our deadlock re backward
>>> compatibility, versioning and RERO:
>>>
>>> Agree that odd numbered versions have stable APIs - basically
>>> adhere
>>> to Commons rules - no breaks within 3.0, 3.1, ..., 3.x... or 5.0,
>>> 5.1... but even-numbered lines can include breaks -
>>
>> I like the proposal to be more lax on compatibility breaking than
>> I ever dreamed of.  ;-)
>>
>>> so 4.0 and 4.1
>>> might not be compatible.
>>
>> Isn't that going to cause JAR hell?
>
> Yes, within the even-numbered branches.  But we would repackage and
> rename at each version cut as we do now.

I don't follow.
Is top-level package renaming in order: e.g.
"org.apache.commons.math4m0" for 4.0
"org.apache.commons.math4m1" for 4.1
?

This would be akin to the "experimental" package idea evoked
in an earlier incarnation of this discussion.

> So the only "hell" would
> be if someone deploys multiple different versions from an
> even-numbered branch.  We would expect them to be less widely
> deployed, so this would be less of an issue.

I don't follow either.
What issue did you intend to solve with the proposal?

My wish is that all (users and developers) can enjoy a timely
official release of the "bleeding edge"; whereas it now seems
to me that you are only talking about branch naming.
[And even-numbered ones will likely not be released because
of JAR hell.]

Or what did I miss?

Gilles

>
> Phil
>>
>> Gilles
>>
>>> We would always maintain both an odd and
>>> even branch - ideally in such a way that when an even numbered line
>>> stabilized it would add a last hurrah of breaks and move to odd.
>>> People wanting stable APIs could just stick with the odd-numbered
>>> lines and [math] developers wanting to experiment with things and
>>> not worry about compatibility could do that in the even-numbered
>>> lines.  In effect, this is sort of what we are doing now in 3.x /
>>> 4.x.
>>>
>>> I know above violates Commons policy if we actually cut releases
>>> from the even-numbered branches - we would have to get agreement
>>> from the Commons PMC that this is OK or somehow label the releases
>>> differently.  Just an idea to get us out of our current bind...
>>>
>>> Phil


---------------------------------------------------------------------
To unsubscribe, e-mail: [hidden email]
For additional commands, e-mail: [hidden email]

Reply | Threaded
Open this post in threaded view
|

Re: [math] Version mgt idea

Luc Maisonobe-2
In reply to this post by sebb-2-2
Le 06/11/2015 18:18, sebb a écrit :

> On 6 November 2015 at 16:17, Phil Steitz <[hidden email]> wrote:
>> Here is an idea that might break our deadlock re backward
>> compatibility, versioning and RERO:
>>
>> Agree that odd numbered versions have stable APIs - basically adhere
>> to Commons rules - no breaks within 3.0, 3.1, ..., 3.x... or 5.0,
>> 5.1... but even-numbered lines can include breaks - so 4.0 and 4.1
>> might not be compatible.  We would always maintain both an odd and
>> even branch - ideally in such a way that when an even numbered line
>> stabilized it would add a last hurrah of breaks and move to odd.
>> People wanting stable APIs could just stick with the odd-numbered
>> lines and [math] developers wanting to experiment with things and
>> not worry about compatibility could do that in the even-numbered
>> lines.  In effect, this is sort of what we are doing now in 3.x / 4.x.
>>
>> I know above violates Commons policy if we actually cut releases
>> from the even-numbered branches - we would have to get agreement
>> from the Commons PMC that this is OK or somehow label the releases
>> differently.  Just an idea to get us out of our current bind...
>
> That would likely cause problems with Maven, which will pick the
> latest release, regardless of whether it is even or odd.

No because we change the artifact coordinates too.
So people who depend on apache.commons.math3 will never see
artifacts from the apache.commons.math4 series.

> [However changing the gid/uid without changing the package name also
> causes problems with Maven.]
>
> Unless it is possible to add a marker to the release such that Maven
> does not consider it when looking for the latest release.
> The only such marker I know of is SNAPSHOT - such releases are not
> added to Maven Central so cannot be accidentally added to the
> classpath.
>
> Maybe there is another marker which could be used that tells Maven not
> to consider that version.
>
> If there is no other such marker, one way to avoid these issues would
> be to not publish the even releases to Maven Central.
> Users would have to download them some other way.
>
> Maybe there is mileage in having a staging repo for Commons as a half-way house.
> Users would need to specifically add the repo to their pom to use the repo.
> [I think I suggested something like this for some Commons Maven
> plugins that were for developer use only]
>
> Note that publishing even-numbered releases to the ASF mirror service
> can also cause problems if the unstable releases are not clearly
> marked.
>
>> Phil
>>
>>
>> ---------------------------------------------------------------------
>> To unsubscribe, e-mail: [hidden email]
>> For additional commands, e-mail: [hidden email]
>>
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: [hidden email]
> For additional commands, e-mail: [hidden email]
>
>


---------------------------------------------------------------------
To unsubscribe, e-mail: [hidden email]
For additional commands, e-mail: [hidden email]

Reply | Threaded
Open this post in threaded view
|

Re: [math] Version mgt idea

Luc Maisonobe-2
In reply to this post by Gilles Sadowski
Le 06/11/2015 18:31, Gilles a écrit :

> Hi.
>
> On Fri, 6 Nov 2015 09:17:18 -0700, Phil Steitz wrote:
>> Here is an idea that might break our deadlock re backward
>> compatibility, versioning and RERO:
>>
>> Agree that odd numbered versions have stable APIs - basically adhere
>> to Commons rules - no breaks within 3.0, 3.1, ..., 3.x... or 5.0,
>> 5.1... but even-numbered lines can include breaks -
>
> I like the proposal to be more lax on compatibility breaking than
> I ever dreamed of.  ;-)

Yeah, good idea!
Go for it.

>
>> so 4.0 and 4.1
>> might not be compatible.
>
> Isn't that going to cause JAR hell?

As long as people are aware there are no compliance implied,
then they know what they have to do when they decide to
use this.

best regards,
Luc

>
> Gilles
>
>> We would always maintain both an odd and
>> even branch - ideally in such a way that when an even numbered line
>> stabilized it would add a last hurrah of breaks and move to odd.
>> People wanting stable APIs could just stick with the odd-numbered
>> lines and [math] developers wanting to experiment with things and
>> not worry about compatibility could do that in the even-numbered
>> lines.  In effect, this is sort of what we are doing now in 3.x / 4.x.
>>
>> I know above violates Commons policy if we actually cut releases
>> from the even-numbered branches - we would have to get agreement
>> from the Commons PMC that this is OK or somehow label the releases
>> differently.  Just an idea to get us out of our current bind...
>>
>> Phil
>>
>>
>> ---------------------------------------------------------------------
>> To unsubscribe, e-mail: [hidden email]
>> For additional commands, e-mail: [hidden email]
>
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: [hidden email]
> For additional commands, e-mail: [hidden email]
>
>


---------------------------------------------------------------------
To unsubscribe, e-mail: [hidden email]
For additional commands, e-mail: [hidden email]

Reply | Threaded
Open this post in threaded view
|

Re: [math] Version mgt idea

Phil Steitz
In reply to this post by Gilles Sadowski
On 11/6/15 11:02 AM, Gilles wrote:

> On Fri, 6 Nov 2015 10:36:51 -0700, Phil Steitz wrote:
>> On 11/6/15 10:31 AM, Gilles wrote:
>>> Hi.
>>>
>>> On Fri, 6 Nov 2015 09:17:18 -0700, Phil Steitz wrote:
>>>> Here is an idea that might break our deadlock re backward
>>>> compatibility, versioning and RERO:
>>>>
>>>> Agree that odd numbered versions have stable APIs - basically
>>>> adhere
>>>> to Commons rules - no breaks within 3.0, 3.1, ..., 3.x... or 5.0,
>>>> 5.1... but even-numbered lines can include breaks -
>>>
>>> I like the proposal to be more lax on compatibility breaking than
>>> I ever dreamed of.  ;-)
>>>
>>>> so 4.0 and 4.1
>>>> might not be compatible.
>>>
>>> Isn't that going to cause JAR hell?
>>
>> Yes, within the even-numbered branches.  But we would repackage and
>> rename at each version cut as we do now.
>
> I don't follow.
> Is top-level package renaming in order: e.g.
> "org.apache.commons.math4m0" for 4.0
> "org.apache.commons.math4m1" for 4.1
> ?

Sorry, I was not clear.  I meant that we would do the repackaging
only at major release, so both above would be .math4


>
> This would be akin to the "experimental" package idea evoked
> in an earlier incarnation of this discussion.
>
>> So the only "hell" would
>> be if someone deploys multiple different versions from an
>> even-numbered branch.  We would expect them to be less widely
>> deployed, so this would be less of an issue.
>
> I don't follow either.
> What issue did you intend to solve with the proposal?

Basically, limit the compat breaks to lines "known" to contain breaks.
>
> My wish is that all (users and developers) can enjoy a timely
> official release of the "bleeding edge"; whereas it now seems
> to me that you are only talking about branch naming.
> [And even-numbered ones will likely not be released because
> of JAR hell.]

That is what I was trying to support.  RERO with constant API change
means you have to release API breaks often.  That's the crux of our
problem with 4.0 now.  If we want it to have a stable API, we need
to finish stabilizing it before we release anything; otherwise we
need to follow immediately with 5, 6 etc.  What I was suggesting is
that we could just advertise the fact that the 4.x API is not
expected to be stable.  Once it stabilizes, it will become 5.x.
This has the advantage that users wanting a stable API know what to
use and we only have to support 2 lines at any given time.

Phil

>
> Or what did I miss?
>
> Gilles
>
>>
>> Phil
>>>
>>> Gilles
>>>
>>>> We would always maintain both an odd and
>>>> even branch - ideally in such a way that when an even numbered
>>>> line
>>>> stabilized it would add a last hurrah of breaks and move to odd.
>>>> People wanting stable APIs could just stick with the odd-numbered
>>>> lines and [math] developers wanting to experiment with things and
>>>> not worry about compatibility could do that in the even-numbered
>>>> lines.  In effect, this is sort of what we are doing now in 3.x /
>>>> 4.x.
>>>>
>>>> I know above violates Commons policy if we actually cut releases
>>>> from the even-numbered branches - we would have to get agreement
>>>> from the Commons PMC that this is OK or somehow label the releases
>>>> differently.  Just an idea to get us out of our current bind...
>>>>
>>>> Phil
>
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: [hidden email]
> For additional commands, e-mail: [hidden email]
>
>



---------------------------------------------------------------------
To unsubscribe, e-mail: [hidden email]
For additional commands, e-mail: [hidden email]

Reply | Threaded
Open this post in threaded view
|

Re: [math] Version mgt idea

Gilles Sadowski
On Fri, 6 Nov 2015 12:21:46 -0700, Phil Steitz wrote:

> On 11/6/15 11:02 AM, Gilles wrote:
>> On Fri, 6 Nov 2015 10:36:51 -0700, Phil Steitz wrote:
>>> On 11/6/15 10:31 AM, Gilles wrote:
>>>> Hi.
>>>>
>>>> On Fri, 6 Nov 2015 09:17:18 -0700, Phil Steitz wrote:
>>>>> Here is an idea that might break our deadlock re backward
>>>>> compatibility, versioning and RERO:
>>>>>
>>>>> Agree that odd numbered versions have stable APIs - basically
>>>>> adhere
>>>>> to Commons rules - no breaks within 3.0, 3.1, ..., 3.x... or 5.0,
>>>>> 5.1... but even-numbered lines can include breaks -
>>>>
>>>> I like the proposal to be more lax on compatibility breaking than
>>>> I ever dreamed of.  ;-)
>>>>
>>>>> so 4.0 and 4.1
>>>>> might not be compatible.
>>>>
>>>> Isn't that going to cause JAR hell?
>>>
>>> Yes, within the even-numbered branches.  But we would repackage and
>>> rename at each version cut as we do now.
>>
>> I don't follow.
>> Is top-level package renaming in order: e.g.
>> "org.apache.commons.math4m0" for 4.0
>> "org.apache.commons.math4m1" for 4.1
>> ?
>
> Sorry, I was not clear.  I meant that we would do the repackaging
> only at major release, so both above would be .math4
>
>
>>
>> This would be akin to the "experimental" package idea evoked
>> in an earlier incarnation of this discussion.
>>
>>> So the only "hell" would
>>> be if someone deploys multiple different versions from an
>>> even-numbered branch.  We would expect them to be less widely
>>> deployed, so this would be less of an issue.
>>
>> I don't follow either.
>> What issue did you intend to solve with the proposal?
>
> Basically, limit the compat breaks to lines "known" to contain
> breaks.
>>
>> My wish is that all (users and developers) can enjoy a timely
>> official release of the "bleeding edge"; whereas it now seems
>> to me that you are only talking about branch naming.
>> [And even-numbered ones will likely not be released because
>> of JAR hell.]
>
> That is what I was trying to support.  RERO with constant API change
> means you have to release API breaks often.  That's the crux of our
> problem with 4.0 now.  If we want it to have a stable API, we need
> to finish stabilizing it before we release anything; otherwise we
> need to follow immediately with 5, 6 etc.  What I was suggesting is
> that we could just advertise the fact that the 4.x API is not
> expected to be stable.  Once it stabilizes, it will become 5.x.
> This has the advantage that users wanting a stable API know what to
> use and we only have to support 2 lines at any given time.

I got it now.  Thanks.

Gilles


---------------------------------------------------------------------
To unsubscribe, e-mail: [hidden email]
For additional commands, e-mail: [hidden email]

Reply | Threaded
Open this post in threaded view
|

Re: [math] Version mgt idea

garydgregory
>
> On Fri, 6 Nov 2015 09:17:18 -0700, Phil Steitz wrote:
>>>>>
>>>>>> Here is an idea that might break our deadlock re backward
>>>>>> compatibility, versioning and RERO:
>>>>>>
>>>>>> Agree that odd numbered versions have stable APIs - basically
>>>>>> adhere
>>>>>> to Commons rules - no breaks within 3.0, 3.1, ..., 3.x... or 5.0,
>>>>>> 5.1... but even-numbered lines can include breaks -
>>>>>>
>>>>> ...

This sounds awfully complicated for my puny human brain.

It's bad enough that I have to remember how each FOSS project treats
versions numbers, but having an exception within a Commons component is
even worse. This is a non-starter for me.

Gary
Reply | Threaded
Open this post in threaded view
|

Re: [math] Version mgt idea

Phil Steitz
On 11/6/15 2:51 PM, Gary Gregory wrote:

>> On Fri, 6 Nov 2015 09:17:18 -0700, Phil Steitz wrote:
>>>>>>> Here is an idea that might break our deadlock re backward
>>>>>>> compatibility, versioning and RERO:
>>>>>>>
>>>>>>> Agree that odd numbered versions have stable APIs - basically
>>>>>>> adhere
>>>>>>> to Commons rules - no breaks within 3.0, 3.1, ..., 3.x... or 5.0,
>>>>>>> 5.1... but even-numbered lines can include breaks -
>>>>>>>
>>>>>> ...
> This sounds awfully complicated for my puny human brain.

How, exactly?  Seems pretty simple to me.  The even-numbered release
lines may have compat breaks; but the odd-numbered do not.
>
> It's bad enough that I have to remember how each FOSS project treats
> versions numbers, but having an exception within a Commons component is
> even worse. This is a non-starter for me.

Do you have any better suggestions?  The problem we are trying to
solve is we can't RERO while sticking to the normal compat rules
without turning major versions all the time, which forces users to
repackage all the time and us to support more versions concurrently
than we have bandwidth to do.

Phil
>
> Gary
>



---------------------------------------------------------------------
To unsubscribe, e-mail: [hidden email]
For additional commands, e-mail: [hidden email]

Reply | Threaded
Open this post in threaded view
|

Re: [math] Version mgt idea

garydgregory
On Fri, Nov 6, 2015 at 3:01 PM, Phil Steitz <[hidden email]> wrote:

> On 11/6/15 2:51 PM, Gary Gregory wrote:
> >> On Fri, 6 Nov 2015 09:17:18 -0700, Phil Steitz wrote:
> >>>>>>> Here is an idea that might break our deadlock re backward
> >>>>>>> compatibility, versioning and RERO:
> >>>>>>>
> >>>>>>> Agree that odd numbered versions have stable APIs - basically
> >>>>>>> adhere
> >>>>>>> to Commons rules - no breaks within 3.0, 3.1, ..., 3.x... or 5.0,
> >>>>>>> 5.1... but even-numbered lines can include breaks -
> >>>>>>>
> >>>>>> ...
> > This sounds awfully complicated for my puny human brain.
>
> How, exactly?  Seems pretty simple to me.  The even-numbered release
> lines may have compat breaks; but the odd-numbered do not.
> >
> > It's bad enough that I have to remember how each FOSS project treats
> > versions numbers, but having an exception within a Commons component is
> > even worse. This is a non-starter for me.
>
> Do you have any better suggestions?  The problem we are trying to
> solve is we can't RERO while sticking to the normal compat rules
> without turning major versions all the time, which forces users to
> repackage all the time and us to support more versions concurrently
> than we have bandwidth to do.
>

I do not see how a different version scheme will determine how many
branches the community supports.

Breaking BC without a package and coord change is a no-go. You have to
think about this jar as a dependency that can be deeply nested in a
software stack. Commons components are such creatures. I unfortunately run
into this more than I'd like: Big FOSS project A depends on B which depends
on C. Then I want to integrate with Project X which depends on Y which
depends on different versions of B and C. Welcome to jar hell if B and C
are not compatible. If B and C follow the rule of break-BC -> new
package/coords, then all is well.

Gary


>
> Phil
> >
> > Gary
> >
>
>
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: [hidden email]
> For additional commands, e-mail: [hidden email]
>
>


--
E-Mail: [hidden email] | [hidden email]
Java Persistence with Hibernate, Second Edition
<http://www.manning.com/bauer3/>
JUnit in Action, Second Edition <http://www.manning.com/tahchiev/>
Spring Batch in Action <http://www.manning.com/templier/>
Blog: http://garygregory.wordpress.com
Home: http://garygregory.com/
Tweet! http://twitter.com/GaryGregory
Reply | Threaded
Open this post in threaded view
|

Re: [math] Version mgt idea

Phil Steitz
On 11/6/15 4:46 PM, Gary Gregory wrote:

> On Fri, Nov 6, 2015 at 3:01 PM, Phil Steitz <[hidden email]> wrote:
>
>> On 11/6/15 2:51 PM, Gary Gregory wrote:
>>>> On Fri, 6 Nov 2015 09:17:18 -0700, Phil Steitz wrote:
>>>>>>>>> Here is an idea that might break our deadlock re backward
>>>>>>>>> compatibility, versioning and RERO:
>>>>>>>>>
>>>>>>>>> Agree that odd numbered versions have stable APIs - basically
>>>>>>>>> adhere
>>>>>>>>> to Commons rules - no breaks within 3.0, 3.1, ..., 3.x... or 5.0,
>>>>>>>>> 5.1... but even-numbered lines can include breaks -
>>>>>>>>>
>>>>>>>> ...
>>> This sounds awfully complicated for my puny human brain.
>> How, exactly?  Seems pretty simple to me.  The even-numbered release
>> lines may have compat breaks; but the odd-numbered do not.
>>> It's bad enough that I have to remember how each FOSS project treats
>>> versions numbers, but having an exception within a Commons component is
>>> even worse. This is a non-starter for me.
>> Do you have any better suggestions?  The problem we are trying to
>> solve is we can't RERO while sticking to the normal compat rules
>> without turning major versions all the time, which forces users to
>> repackage all the time and us to support more versions concurrently
>> than we have bandwidth to do.
>>
> I do not see how a different version scheme will determine how many
> branches the community supports.

If we just keep one 4.x branch that keeps cutting (possibly
incompatible) releases, that is just one line, one branch.  If we
have to cut 4.1, 4.2, 4.3 as 4, 5, 6 instead and we don't allow any
compat breaks, we end up having to maintain and release 4.0.1,
5.0.1, 6.0.1 instead of just 4.3.1, for example, or we just strand
the 4, 5 users in terms of bug fixes as we move on to 6.
>
> Breaking BC without a package and coord change is a no-go.

We have done this before and we will probably do it again - and more
if we have to don't separate out an unstable line.
>  You have to
> think about this jar as a dependency that can be deeply nested in a
> software stack. Commons components are such creatures. I unfortunately run
> into this more than I'd like: Big FOSS project A depends on B which depends
> on C. Then I want to integrate with Project X which depends on Y which
> depends on different versions of B and C. Welcome to jar hell if B and C
> are not compatible. If B and C follow the rule of break-BC -> new
> package/coords, then all is well.

The mitigation here is that we would not expect the even-numbered
releases to be deployed widely.

Phil

>
> Gary
>
>
>> Phil
>>> Gary
>>>
>>
>>
>> ---------------------------------------------------------------------
>> To unsubscribe, e-mail: [hidden email]
>> For additional commands, e-mail: [hidden email]
>>
>>
>



---------------------------------------------------------------------
To unsubscribe, e-mail: [hidden email]
For additional commands, e-mail: [hidden email]

Reply | Threaded
Open this post in threaded view
|

Re: [math] Version mgt idea

Gilles Sadowski
On Fri, 6 Nov 2015 17:02:01 -0700, Phil Steitz wrote:

> On 11/6/15 4:46 PM, Gary Gregory wrote:
>> On Fri, Nov 6, 2015 at 3:01 PM, Phil Steitz <[hidden email]>
>> wrote:
>>
>>> On 11/6/15 2:51 PM, Gary Gregory wrote:
>>>>> On Fri, 6 Nov 2015 09:17:18 -0700, Phil Steitz wrote:
>>>>>>>>>> Here is an idea that might break our deadlock re backward
>>>>>>>>>> compatibility, versioning and RERO:
>>>>>>>>>>
>>>>>>>>>> Agree that odd numbered versions have stable APIs -
>>>>>>>>>> basically
>>>>>>>>>> adhere
>>>>>>>>>> to Commons rules - no breaks within 3.0, 3.1, ..., 3.x... or
>>>>>>>>>> 5.0,
>>>>>>>>>> 5.1... but even-numbered lines can include breaks -
>>>>>>>>>>
>>>>>>>>> ...
>>>> This sounds awfully complicated for my puny human brain.
>>> How, exactly?  Seems pretty simple to me.  The even-numbered
>>> release
>>> lines may have compat breaks; but the odd-numbered do not.
>>>> It's bad enough that I have to remember how each FOSS project
>>>> treats
>>>> versions numbers, but having an exception within a Commons
>>>> component is
>>>> even worse. This is a non-starter for me.
>>> Do you have any better suggestions?  The problem we are trying to
>>> solve is we can't RERO while sticking to the normal compat rules
>>> without turning major versions all the time, which forces users to
>>> repackage all the time and us to support more versions concurrently
>>> than we have bandwidth to do.
>>>
>> I do not see how a different version scheme will determine how many
>> branches the community supports.
>
> If we just keep one 4.x branch that keeps cutting (possibly
> incompatible) releases, that is just one line, one branch.  If we
> have to cut 4.1, 4.2, 4.3 as 4, 5, 6 instead and we don't allow any
> compat breaks, we end up having to maintain and release 4.0.1,
> 5.0.1, 6.0.1 instead of just 4.3.1, for example, or we just strand
> the 4, 5 users in terms of bug fixes as we move on to 6.
>>
>> Breaking BC without a package and coord change is a no-go.
>
> We have done this before and we will probably do it again - and more
> if we have to don't separate out an unstable line.
>>  You have to
>> think about this jar as a dependency that can be deeply nested in a
>> software stack. Commons components are such creatures. I
>> unfortunately run
>> into this more than I'd like: Big FOSS project A depends on B which
>> depends
>> on C. Then I want to integrate with Project X which depends on Y
>> which
>> depends on different versions of B and C. Welcome to jar hell if B
>> and C
>> are not compatible. If B and C follow the rule of break-BC -> new
>> package/coords, then all is well.
>
> The mitigation here is that we would not expect the even-numbered
> releases to be deployed widely.

Or we can prevent any potential JAR hell by changing the top-level
package name for every release in the unstable series:

4.0 -> org.apache.commons.math4u0
4.1 -> org.apache.commons.math4u1

This would also offer the possibility to write a code using
classes from several unstable releases (e.g. to compare their
performance).

Gilles


---------------------------------------------------------------------
To unsubscribe, e-mail: [hidden email]
For additional commands, e-mail: [hidden email]

Reply | Threaded
Open this post in threaded view
|

Re: [math] Version mgt idea

garydgregory
In reply to this post by Phil Steitz
On Fri, Nov 6, 2015 at 4:02 PM, Phil Steitz <[hidden email]> wrote:

> On 11/6/15 4:46 PM, Gary Gregory wrote:
> > On Fri, Nov 6, 2015 at 3:01 PM, Phil Steitz <[hidden email]>
> wrote:
> >
> >> On 11/6/15 2:51 PM, Gary Gregory wrote:
> >>>> On Fri, 6 Nov 2015 09:17:18 -0700, Phil Steitz wrote:
> >>>>>>>>> Here is an idea that might break our deadlock re backward
> >>>>>>>>> compatibility, versioning and RERO:
> >>>>>>>>>
> >>>>>>>>> Agree that odd numbered versions have stable APIs - basically
> >>>>>>>>> adhere
> >>>>>>>>> to Commons rules - no breaks within 3.0, 3.1, ..., 3.x... or 5.0,
> >>>>>>>>> 5.1... but even-numbered lines can include breaks -
> >>>>>>>>>
> >>>>>>>> ...
> >>> This sounds awfully complicated for my puny human brain.
> >> How, exactly?  Seems pretty simple to me.  The even-numbered release
> >> lines may have compat breaks; but the odd-numbered do not.
> >>> It's bad enough that I have to remember how each FOSS project treats
> >>> versions numbers, but having an exception within a Commons component is
> >>> even worse. This is a non-starter for me.
> >> Do you have any better suggestions?  The problem we are trying to
> >> solve is we can't RERO while sticking to the normal compat rules
> >> without turning major versions all the time, which forces users to
> >> repackage all the time and us to support more versions concurrently
> >> than we have bandwidth to do.
> >>
> > I do not see how a different version scheme will determine how many
> > branches the community supports.
>
> If we just keep one 4.x branch that keeps cutting (possibly
> incompatible) releases, that is just one line, one branch.  If we
> have to cut 4.1, 4.2, 4.3 as 4, 5, 6 instead and we don't allow any
> compat breaks, we end up having to maintain and release 4.0.1,
> 5.0.1, 6.0.1 instead of just 4.3.1, for example, or we just strand
> the 4, 5 users in terms of bug fixes as we move on to 6.
> >
> > Breaking BC without a package and coord change is a no-go.
>
> We have done this before and we will probably do it again - and more
> if we have to don't separate out an unstable line.
> >  You have to
> > think about this jar as a dependency that can be deeply nested in a
> > software stack. Commons components are such creatures. I unfortunately
> run
> > into this more than I'd like: Big FOSS project A depends on B which
> depends
> > on C. Then I want to integrate with Project X which depends on Y which
> > depends on different versions of B and C. Welcome to jar hell if B and C
> > are not compatible. If B and C follow the rule of break-BC -> new
> > package/coords, then all is well.
>
> The mitigation here is that we would not expect the even-numbered
> releases to be deployed widely.
>

Respectfully Phil, my point is that while this might be true, it is in
practice irrelevant. We cannot control the spread of our jars and their
usage, hence the importance of BC.

Gary

>
> Phil
> >
> > Gary
> >
> >
> >> Phil
> >>> Gary
> >>>
> >>
> >>
> >> ---------------------------------------------------------------------
> >> To unsubscribe, e-mail: [hidden email]
> >> For additional commands, e-mail: [hidden email]
> >>
> >>
> >
>
>
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: [hidden email]
> For additional commands, e-mail: [hidden email]
>
>


--
E-Mail: [hidden email] | [hidden email]
Java Persistence with Hibernate, Second Edition
<http://www.manning.com/bauer3/>
JUnit in Action, Second Edition <http://www.manning.com/tahchiev/>
Spring Batch in Action <http://www.manning.com/templier/>
Blog: http://garygregory.wordpress.com
Home: http://garygregory.com/
Tweet! http://twitter.com/GaryGregory
Reply | Threaded
Open this post in threaded view
|

Re: [math] Version mgt idea

garydgregory
In reply to this post by Gilles Sadowski
On Fri, Nov 6, 2015 at 4:42 PM, Gilles <[hidden email]> wrote:

> On Fri, 6 Nov 2015 17:02:01 -0700, Phil Steitz wrote:
>
>> On 11/6/15 4:46 PM, Gary Gregory wrote:
>>
>>> On Fri, Nov 6, 2015 at 3:01 PM, Phil Steitz <[hidden email]>
>>> wrote:
>>>
>>> On 11/6/15 2:51 PM, Gary Gregory wrote:
>>>>
>>>>> On Fri, 6 Nov 2015 09:17:18 -0700, Phil Steitz wrote:
>>>>>>
>>>>>>> Here is an idea that might break our deadlock re backward
>>>>>>>>>>> compatibility, versioning and RERO:
>>>>>>>>>>>
>>>>>>>>>>> Agree that odd numbered versions have stable APIs - basically
>>>>>>>>>>> adhere
>>>>>>>>>>> to Commons rules - no breaks within 3.0, 3.1, ..., 3.x... or 5.0,
>>>>>>>>>>> 5.1... but even-numbered lines can include breaks -
>>>>>>>>>>>
>>>>>>>>>>> ...
>>>>>>>>>>
>>>>>>>>> This sounds awfully complicated for my puny human brain.
>>>>>
>>>> How, exactly?  Seems pretty simple to me.  The even-numbered release
>>>> lines may have compat breaks; but the odd-numbered do not.
>>>>
>>>>> It's bad enough that I have to remember how each FOSS project treats
>>>>> versions numbers, but having an exception within a Commons component is
>>>>> even worse. This is a non-starter for me.
>>>>>
>>>> Do you have any better suggestions?  The problem we are trying to
>>>> solve is we can't RERO while sticking to the normal compat rules
>>>> without turning major versions all the time, which forces users to
>>>> repackage all the time and us to support more versions concurrently
>>>> than we have bandwidth to do.
>>>>
>>>> I do not see how a different version scheme will determine how many
>>> branches the community supports.
>>>
>>
>> If we just keep one 4.x branch that keeps cutting (possibly
>> incompatible) releases, that is just one line, one branch.  If we
>> have to cut 4.1, 4.2, 4.3 as 4, 5, 6 instead and we don't allow any
>> compat breaks, we end up having to maintain and release 4.0.1,
>> 5.0.1, 6.0.1 instead of just 4.3.1, for example, or we just strand
>> the 4, 5 users in terms of bug fixes as we move on to 6.
>>
>>>
>>> Breaking BC without a package and coord change is a no-go.
>>>
>>
>> We have done this before and we will probably do it again - and more
>> if we have to don't separate out an unstable line.
>>
>>>  You have to
>>> think about this jar as a dependency that can be deeply nested in a
>>> software stack. Commons components are such creatures. I unfortunately
>>> run
>>> into this more than I'd like: Big FOSS project A depends on B which
>>> depends
>>> on C. Then I want to integrate with Project X which depends on Y which
>>> depends on different versions of B and C. Welcome to jar hell if B and C
>>> are not compatible. If B and C follow the rule of break-BC -> new
>>> package/coords, then all is well.
>>>
>>
>> The mitigation here is that we would not expect the even-numbered
>> releases to be deployed widely.
>>
>
> Or we can prevent any potential JAR hell by changing the top-level
> package name for every release in the unstable series:
>
> 4.0 -> org.apache.commons.math4u0
> 4.1 -> org.apache.commons.math4u1
>
> This would also offer the possibility to write a code using
> classes from several unstable releases (e.g. to compare their
> performance).


Pardon my daftness, but if you change package/coord names, what are you
gaining compared to the usual major version changes?

Gary



>
> Gilles
>
>
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: [hidden email]
> For additional commands, e-mail: [hidden email]
>
>


--
E-Mail: [hidden email] | [hidden email]
Java Persistence with Hibernate, Second Edition
<http://www.manning.com/bauer3/>
JUnit in Action, Second Edition <http://www.manning.com/tahchiev/>
Spring Batch in Action <http://www.manning.com/templier/>
Blog: http://garygregory.wordpress.com
Home: http://garygregory.com/
Tweet! http://twitter.com/GaryGregory
Reply | Threaded
Open this post in threaded view
|

Re: [math] Version mgt idea

Gilles Sadowski
On Fri, 6 Nov 2015 16:53:00 -0800, Gary Gregory wrote:

> On Fri, Nov 6, 2015 at 4:42 PM, Gilles <[hidden email]>
> wrote:
>
>> On Fri, 6 Nov 2015 17:02:01 -0700, Phil Steitz wrote:
>>
>>> On 11/6/15 4:46 PM, Gary Gregory wrote:
>>>
>>>> On Fri, Nov 6, 2015 at 3:01 PM, Phil Steitz
>>>> <[hidden email]>
>>>> wrote:
>>>>
>>>> On 11/6/15 2:51 PM, Gary Gregory wrote:
>>>>>
>>>>>> On Fri, 6 Nov 2015 09:17:18 -0700, Phil Steitz wrote:
>>>>>>>
>>>>>>>> Here is an idea that might break our deadlock re backward
>>>>>>>>>>>> compatibility, versioning and RERO:
>>>>>>>>>>>>
>>>>>>>>>>>> Agree that odd numbered versions have stable APIs -
>>>>>>>>>>>> basically
>>>>>>>>>>>> adhere
>>>>>>>>>>>> to Commons rules - no breaks within 3.0, 3.1, ..., 3.x...
>>>>>>>>>>>> or 5.0,
>>>>>>>>>>>> 5.1... but even-numbered lines can include breaks -
>>>>>>>>>>>>
>>>>>>>>>>>> ...
>>>>>>>>>>>
>>>>>>>>>> This sounds awfully complicated for my puny human brain.
>>>>>>
>>>>> How, exactly?  Seems pretty simple to me.  The even-numbered
>>>>> release
>>>>> lines may have compat breaks; but the odd-numbered do not.
>>>>>
>>>>>> It's bad enough that I have to remember how each FOSS project
>>>>>> treats
>>>>>> versions numbers, but having an exception within a Commons
>>>>>> component is
>>>>>> even worse. This is a non-starter for me.
>>>>>>
>>>>> Do you have any better suggestions?  The problem we are trying to
>>>>> solve is we can't RERO while sticking to the normal compat rules
>>>>> without turning major versions all the time, which forces users
>>>>> to
>>>>> repackage all the time and us to support more versions
>>>>> concurrently
>>>>> than we have bandwidth to do.
>>>>>
>>>>> I do not see how a different version scheme will determine how
>>>>> many
>>>> branches the community supports.
>>>>
>>>
>>> If we just keep one 4.x branch that keeps cutting (possibly
>>> incompatible) releases, that is just one line, one branch.  If we
>>> have to cut 4.1, 4.2, 4.3 as 4, 5, 6 instead and we don't allow any
>>> compat breaks, we end up having to maintain and release 4.0.1,
>>> 5.0.1, 6.0.1 instead of just 4.3.1, for example, or we just strand
>>> the 4, 5 users in terms of bug fixes as we move on to 6.
>>>
>>>>
>>>> Breaking BC without a package and coord change is a no-go.
>>>>
>>>
>>> We have done this before and we will probably do it again - and
>>> more
>>> if we have to don't separate out an unstable line.
>>>
>>>>  You have to
>>>> think about this jar as a dependency that can be deeply nested in
>>>> a
>>>> software stack. Commons components are such creatures. I
>>>> unfortunately
>>>> run
>>>> into this more than I'd like: Big FOSS project A depends on B
>>>> which
>>>> depends
>>>> on C. Then I want to integrate with Project X which depends on Y
>>>> which
>>>> depends on different versions of B and C. Welcome to jar hell if B
>>>> and C
>>>> are not compatible. If B and C follow the rule of break-BC -> new
>>>> package/coords, then all is well.
>>>>
>>>
>>> The mitigation here is that we would not expect the even-numbered
>>> releases to be deployed widely.
>>>
>>
>> Or we can prevent any potential JAR hell by changing the top-level
>> package name for every release in the unstable series:
>>
>> 4.0 -> org.apache.commons.math4u0
>> 4.1 -> org.apache.commons.math4u1
>>
>> This would also offer the possibility to write a code using
>> classes from several unstable releases (e.g. to compare their
>> performance).
>
>
> Pardon my daftness, but if you change package/coord names, what are
> you
> gaining compared to the usual major version changes?
>

Objectively, nothing...
But IIUC Phil's paragraph, above, using a dual numbering would
highlight the "unstable" nature of some releases: the whole 4.x
series would be known (in advance) to be unstable, while if we
release a usual 4.0, some user might assume it to be "stable",
only to see it abandoned a few months later when 5.0 is released.
Moreover, at each major number bump, we'd have to advertize (in
words) whether that release is stable or not, while with the dual
numbering the warning would be carried by number in the package
name.

Maybe it's only a psychological trick.  But if it can help to
release more often, it would have led to a real improvement! :)

Gilles



---------------------------------------------------------------------
To unsubscribe, e-mail: [hidden email]
For additional commands, e-mail: [hidden email]

Reply | Threaded
Open this post in threaded view
|

Re: [math] Version mgt idea

Emmanuel Bourg-3
In reply to this post by garydgregory
A roughly equivalent alternative would be to release beta artifacts
until the API stabilizes and use a different base package and different
Maven coordinates for each iteration.

For example, commons-math 4.0-beta1 is released with the
org.apache.commons:commons-math4-beta1 coordinates and the classes
living in the org.apache.commons.math4.beta1 package. Once we are happy
with the state of the API we release org.apache.commons:commons-math4
with the org.apache.commons.math4 base package and we stop breaking the
binary compatibility.

In my opinion the "beta" qualifier better conveys the unstable nature of
the API than an arbitrary convention like "the whole 4.x line is
unstable". This is a concept people are already familiar with.

Emmanuel Bourg


---------------------------------------------------------------------
To unsubscribe, e-mail: [hidden email]
For additional commands, e-mail: [hidden email]

Reply | Threaded
Open this post in threaded view
|

Re: [math] Version mgt idea

James Carman
In reply to this post by Phil Steitz
On Fri, Nov 6, 2015 at 11:17 AM Phil Steitz <[hidden email]> wrote:

> Here is an idea that might break our deadlock re backward
> compatibility, versioning and RERO:
>
> Agree that odd numbered versions have stable APIs - basically adhere
> to Commons rules - no breaks within 3.0, 3.1, ..., 3.x... or 5.0,
> 5.1... but even-numbered lines can include breaks - so 4.0 and 4.1
> might not be compatible.  We would always maintain both an odd and
> even branch - ideally in such a way that when an even numbered line
> stabilized it would add a last hurrah of breaks and move to odd.
> People wanting stable APIs could just stick with the odd-numbered
> lines and [math] developers wanting to experiment with things and
> not worry about compatibility could do that in the even-numbered
> lines.  In effect, this is sort of what we are doing now in 3.x / 4.x.
>
> I know above violates Commons policy if we actually cut releases
> from the even-numbered branches - we would have to get agreement
> from the Commons PMC that this is OK or somehow label the releases
> differently.  Just an idea to get us out of our current bind...
>
>
I'm -1 on this idea.  I don't like the idea of causing compatibility breaks
within a major revision, no matter what.  We should follow semantic
versioning guidelines wrt compatibility breaks.  If not, you're going to
run into trouble in places like OSGi, which assumes you're following those
rules.
12