Re: Are snapshots available in some Maven repo?

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

Re: Are snapshots available in some Maven repo?

Mickael Istria-2
On Wednesday, January 9, 2019, Robert Scholte <[hidden email]> wrote:

> I've just deployed the SNAPSHOTs from current master to
> repository.apache.org
>

Thanks a lot! Could it be done continuously? Technicallu, a CI job.checking
the repo could tale care of doing it soon after each change without further
human effort.


--
Mickael Istria
Eclipse IDE <https://www.eclipse.org/downloads/eclipse-packages/>
developer, for Red Hat Developers <https://developers.redhat.com/>
Reply | Threaded
Open this post in threaded view
|

Re: Are snapshots available in some Maven repo?

Enrico Olivelli
We should only have to setup a CI job which does
mvn deploy

AFAIK ASF Jenkins are allowed to push snapshots

Enrico

Il gio 10 gen 2019, 00:34 Mickael Istria <[hidden email]> ha scritto:

> On Wednesday, January 9, 2019, Robert Scholte <[hidden email]>
> wrote:
>
> > I've just deployed the SNAPSHOTs from current master to
> > repository.apache.org
> >
>
> Thanks a lot! Could it be done continuously? Technicallu, a CI job.checking
> the repo could tale care of doing it soon after each change without further
> human effort.
>
>
> --
> Mickael Istria
> Eclipse IDE <https://www.eclipse.org/downloads/eclipse-packages/>
> developer, for Red Hat Developers <https://developers.redhat.com/>
>
--


-- Enrico Olivelli
Reply | Threaded
Open this post in threaded view
|

Re: Are snapshots available in some Maven repo?

stephenconnolly
The question is more what our policy is.

In my opinion you need one of two policies:

* snapshots are deployed manually
* snapshots are deployed automatically from a specific branch

We have used manual as our policy. If we change that’s fine, but we should
stop manual deployment and clarify the deployment frequency so that people
can set their update policy in their settings.xml appropriately.

A vote would be required imho... and a discuss beforehand.

On Thu 10 Jan 2019 at 06:18, Enrico Olivelli <[hidden email]> wrote:

> We should only have to setup a CI job which does
> mvn deploy
>
> AFAIK ASF Jenkins are allowed to push snapshots
>
> Enrico
>
> Il gio 10 gen 2019, 00:34 Mickael Istria <[hidden email]> ha scritto:
>
> > On Wednesday, January 9, 2019, Robert Scholte <[hidden email]>
> > wrote:
> >
> > > I've just deployed the SNAPSHOTs from current master to
> > > repository.apache.org
> > >
> >
> > Thanks a lot! Could it be done continuously? Technicallu, a CI
> job.checking
> > the repo could tale care of doing it soon after each change without
> further
> > human effort.
> >
> >
> > --
> > Mickael Istria
> > Eclipse IDE <https://www.eclipse.org/downloads/eclipse-packages/>
> > developer, for Red Hat Developers <https://developers.redhat.com/>
> >
> --
>
>
> -- Enrico Olivelli
>
--
Sent from my phone
Reply | Threaded
Open this post in threaded view
|

Re: Are snapshots available in some Maven repo?

rfscholte
Now that we use branches actively, it is way more important where a  
SNAPSHOT is coming from.
I've seen it too often that people thought they were testing with a  
version from a specific branch, but that it was redeployed with another  
version.
The first steps are done to rewrite the pom-file during publication, and  
that might be exactly what we need here: when we're on a branch, the  
branchname should be added to the version.
Call it work in progress for better support.

thanks,
Robert

On Thu, 10 Jan 2019 10:50:57 +0100, Stephen Connolly  
<[hidden email]> wrote:

> On Thu, 10 Jan 2019 at 09:11, Mickael Istria <[hidden email]> wrote:
>
>> Hi,
>>
>> Eclipse m2e, as a consumer of Maven as a library, would love to see the
>> latest HEAD from master published automatically as SNAPSHOTs soon after
>> every change is made. This seems like a requirement to enable continuous
>> feedback from m2e to Maven.
>> Publishing some other branches or commits doesn't seem interesting to
>> Eclipse m2e at the moment.
>>
>
> My point is that we need a clearly defined policy as to exactly what
> branch(es) are automatically pushed and when.
>
> That policy could be *one* of:
> * "master" if there are SCM changes detected on Sunday at 00:00 UTC
> * "master" if there are SCM changes detected at 00:00 UTC
> * "master" as soon as SCM changes are detected
> * "integration" as soon as SCM changes are detected
> * etc
>
> We just need to be crystal clear exactly what policy we are following.
> Right now the policy we are following is:
>
> * Manually from whatever branch is appropriate and only when there is a
> request for them to be published.
>
> There are good and valid reasons for any of these policies (or indeed  
> other
> alternative policies). The problems occur if you try to mix policy.
>
> For example, if I as a developer know that snapshots could be published  
> at
> any time then I can choose between `always` and `never` as the update
> policy depending on whether I want to focus on diagnosing issues on a
> consistent base or discovering issues on shifting sand.
>
> If I know that snapshots will only ever be published once a week then I  
> can
> leave the update policy at the default of daily because I'll need to
> rebuild my mental context on the next week anyway and it's only the first
> build on Monday's that will have the time kill.
>
> So the thing that we need to do is decide if we want to change our  
> policy,
> if yes, then decide what we want to change it to *and* both implement it
> consistently and publish it clearly.
>
> Finally we have the ASF legal requirements that mean we need to clarify  
> to
> all that -SNAPSHOTs are not actually releases and are only made available
> for validation prior to VOTE threads... because currently -SNAPSHOTs are
> rare this is not a big issue... if we have the CI pushing -SNAPSHOTs at a
> regular cadence the risk of *users* starting to use -SNAPSHOTs as  
> releases
> rises so we would need to find ways to mitigate such risks
>
>
>>
>> Cheers,

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

Reply | Threaded
Open this post in threaded view
|

Re: Are snapshots available in some Maven repo?

Enrico Olivelli
+1 for master and nightly, timezone is not so important

Enrico

Il gio 10 gen 2019, 21:44 Stephen Connolly <[hidden email]>
ha scritto:

> I say if we are doing automatic snapshot deployment, master branch only and
> either weekly or daily or on every commit... pick one for all our projects.
>
> Other branches would just confuse things so keep them as manual deployment
> and if doing that require the branch name included in the version number.
>
> On Thu 10 Jan 2019 at 20:14, Olivier Lamy <[hidden email]> wrote:
>
> > We already had this discussion so many times and it always turns to non
> > ended complicated discussion with very complicated cases which NEVER
> happen
> > here....
> > So the most important is: We must take it easy and avoid
> overthinking!.....
> > +1 for every night a master snapshot or every master changes.
> > but please keep it simple!!!!
> >
> >
> > On Fri, 11 Jan 2019 at 06:34, Dan Tran <[hidden email]> wrote:
> >
> > > how about just deploy nightly snapshot of active 'main' branch ( ie
> > > master)?
> > >
> > > -D
> > >
> > >
> > > On Thu, Jan 10, 2019 at 9:03 AM Robert Scholte <[hidden email]>
> > > wrote:
> > >
> > > > Now that we use branches actively, it is way more important where a
> > > > SNAPSHOT is coming from.
> > > > I've seen it too often that people thought they were testing with a
> > > > version from a specific branch, but that it was redeployed with
> another
> > > > version.
> > > > The first steps are done to rewrite the pom-file during publication,
> > and
> > > > that might be exactly what we need here: when we're on a branch, the
> > > > branchname should be added to the version.
> > > > Call it work in progress for better support.
> > > >
> > > > thanks,
> > > > Robert
> > > >
> > > > On Thu, 10 Jan 2019 10:50:57 +0100, Stephen Connolly
> > > > <[hidden email]> wrote:
> > > >
> > > > > On Thu, 10 Jan 2019 at 09:11, Mickael Istria <[hidden email]>
> > > wrote:
> > > > >
> > > > >> Hi,
> > > > >>
> > > > >> Eclipse m2e, as a consumer of Maven as a library, would love to
> see
> > > the
> > > > >> latest HEAD from master published automatically as SNAPSHOTs soon
> > > after
> > > > >> every change is made. This seems like a requirement to enable
> > > continuous
> > > > >> feedback from m2e to Maven.
> > > > >> Publishing some other branches or commits doesn't seem interesting
> > to
> > > > >> Eclipse m2e at the moment.
> > > > >>
> > > > >
> > > > > My point is that we need a clearly defined policy as to exactly
> what
> > > > > branch(es) are automatically pushed and when.
> > > > >
> > > > > That policy could be *one* of:
> > > > > * "master" if there are SCM changes detected on Sunday at 00:00 UTC
> > > > > * "master" if there are SCM changes detected at 00:00 UTC
> > > > > * "master" as soon as SCM changes are detected
> > > > > * "integration" as soon as SCM changes are detected
> > > > > * etc
> > > > >
> > > > > We just need to be crystal clear exactly what policy we are
> > following.
> > > > > Right now the policy we are following is:
> > > > >
> > > > > * Manually from whatever branch is appropriate and only when there
> > is a
> > > > > request for them to be published.
> > > > >
> > > > > There are good and valid reasons for any of these policies (or
> indeed
> > > > > other
> > > > > alternative policies). The problems occur if you try to mix policy.
> > > > >
> > > > > For example, if I as a developer know that snapshots could be
> > published
> > > > > at
> > > > > any time then I can choose between `always` and `never` as the
> update
> > > > > policy depending on whether I want to focus on diagnosing issues
> on a
> > > > > consistent base or discovering issues on shifting sand.
> > > > >
> > > > > If I know that snapshots will only ever be published once a week
> > then I
> > > > > can
> > > > > leave the update policy at the default of daily because I'll need
> to
> > > > > rebuild my mental context on the next week anyway and it's only the
> > > first
> > > > > build on Monday's that will have the time kill.
> > > > >
> > > > > So the thing that we need to do is decide if we want to change our
> > > > > policy,
> > > > > if yes, then decide what we want to change it to *and* both
> implement
> > > it
> > > > > consistently and publish it clearly.
> > > > >
> > > > > Finally we have the ASF legal requirements that mean we need to
> > clarify
> > > > > to
> > > > > all that -SNAPSHOTs are not actually releases and are only made
> > > available
> > > > > for validation prior to VOTE threads... because currently
> -SNAPSHOTs
> > > are
> > > > > rare this is not a big issue... if we have the CI pushing
> -SNAPSHOTs
> > > at a
> > > > > regular cadence the risk of *users* starting to use -SNAPSHOTs as
> > > > > releases
> > > > > rises so we would need to find ways to mitigate such risks
> > > > >
> > > > >
> > > > >>
> > > > >> Cheers,
> > > >
> > > > ---------------------------------------------------------------------
> > > > To unsubscribe, e-mail: [hidden email]
> > > > For additional commands, e-mail: [hidden email]
> > > >
> > > >
> > >
> >
> >
> > --
> > Olivier Lamy
> > http://twitter.com/olamy | http://linkedin.com/in/olamy
> >
> --
> Sent from my phone
>
--


-- Enrico Olivelli
Reply | Threaded
Open this post in threaded view
|

Re: Are snapshots available in some Maven repo?

Mickael Istria-2
Hi,

master seems by far enough to me, and I think support for specific other
branches should be considered in another future step if it happens to be
requested one day.
About nightly, I think it's usually not such a good rule. I wrote about it
some time ago when I started to deal with release engineering for a bunch
of Eclipse projects:
https://mickaelistria.wordpress.com/2011/12/07/call-a-spade-a-spade-and-a-nightly-a-snapshot/
. My recommandation would be to use the "Poll SCM" Jenkins feature
configured to check SCM @hourly or so. This should overall produce less
builds than a nightly (leading to less things to re-download and re-try for
clients without actual payload change, so it's more energy-efficient and
easier for testing), and also gives more agility by reducing the feedback
loop.

Cheers,
Reply | Threaded
Open this post in threaded view
|

Re: Are snapshots available in some Maven repo?

Mickael Istria-2
Hi again,

The (hopefully) final patch to make save GBs of RAM in m2e was merged in
Maven earlier today.
I'd like m2e CI builds to be able to test against it. Can someone please
publish the current master as SNAPSHOT ?

Cheers,