You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@marmotta.apache.org by Sergio Fernández <wi...@apache.org> on 2013/04/09 08:40:31 UTC

branching workflow

Hi,

this is something we already did in the previous release candidate, but 
I think is interesting to describe with detail.

According what we discusses back in January, we agreed on using a 
branching workflow, as described at [1].

Then, when the RC5 vote started, we opened a new develop branch where 
continue working on all features would not go in the release, but for 
3.1.0-incubating-SNAPSHOT. We kept the master, and there I was fixing of 
the issues with the release itself.

Some of the changes pushed to develop where actually bug fixes that may 
be interesting to release; so I merged those concrete commits (cherry 
pick) before normally proceeding with the release process [2].

So, what I just did this morning is to merge all changes we did in 
master into develop. This allows us to continue working on develop while 
we are in the voting process, and manage back in case it fails.

Please, do not hesitate to ask any question about the workflow. I think 
the discussion would be quite rewarding for all of us.

Kind regards,

[1] http://marmotta.incubator.apache.org/development.html
[2] http://wiki.apache.org/marmotta/ReleaseProcess

-- 
Sergio Fernández

Re: branching workflow

Posted by Jakob Frank <ja...@apache.org>.
No git-expert here either - but i just observed the following:

jakob@X201:~/workspaces/marmotta/marmotta$ git fetch 
...
   a12b381..8e87a83  master     -> origin/master
 * [new branch]      develop    -> origin/develop
>From https://git-wip-us.apache.org/repos/asf/incubator-marmotta
 * [new tag]         check-license-notice -> check-license-notice

jakob@X201:~/workspaces/marmotta/marmotta$ git fetch --tags 
>From https://git-wip-us.apache.org/repos/asf/incubator-marmotta
 - [tag update]      3.0.0-incubating -> 3.0.0-incubating


So my interpretation:
"git fetch" retrieves the branches and *NEW* tags
"git fetch --tags" also checks for *UPDATED* tags

Can someone verify?

Best
Jakob

On Fri, 2013-04-12 at 09:00 +0100, Andy Seaborne wrote:
> On 12/04/13 07:55, Sergio Fernández wrote:
> > Any expert on git in the room? I thin we have a problem understanding
> > the tagging system.
> >
> > According the documentation: when "git fetch" is run, unless told
> > otherwise by an explicit --no-tags, it automatically fetches tags that
> > point at new commits on branches you fetch.
> >
> > But this does not work for me, and I need to run "git fetch --tags" to
> > actually retrieve the tags. Same for pushing.
> >
> > What this concept error is causing is: we removed some old tags, but not
> > all clones did it, so on new push we are getting back those tags again.
> >
> > I'd really appreciate if someone could bring a proper explanation to
> > such behaviour.
> >
> > Thanks!
> 
> (tries it out going RC6 to RC7 ...)
> 
> Do you mean specifically for "3.0.0-incubating"?
> 
> The reuse of the same name to move from RC(n) to RC(n+1) has given me 
> some sync issues.
> 
> Yes - I needed "--tags".
> 
> 
> git fetch
> blah... blah
> 
> git checkout 3.0.0-incubating
> HEAD is now at a3f52e9... [maven-release-plugin] prepare release 
> 3.0.0-incubating
> 
> 
> git fetch --tags
> 
>  From https://git-wip-us.apache.org/repos/asf/incubator-marmotta
>   - [tag update]      3.0.0-incubating -> 3.0.0-incubating
> 
> git checkout 3.0.0-incubating
> 
> Previous HEAD position was a3f52e9... [maven-release-plugin] prepare 
> release 3.0.0-incubating
> HEAD is now at 63762a3... [maven-release-plugin] prepare release 
> 3.0.0-incubating
> 
> 	 Andy ("not a git expert")
> 
> >
> > On 09/04/13 08:40, Sergio Fernández wrote:
> >> Hi,
> >>
> >> this is something we already did in the previous release candidate, but
> >> I think is interesting to describe with detail.
> >>
> >> According what we discusses back in January, we agreed on using a
> >> branching workflow, as described at [1].
> >>
> >> Then, when the RC5 vote started, we opened a new develop branch where
> >> continue working on all features would not go in the release, but for
> >> 3.1.0-incubating-SNAPSHOT. We kept the master, and there I was fixing of
> >> the issues with the release itself.
> >>
> >> Some of the changes pushed to develop where actually bug fixes that may
> >> be interesting to release; so I merged those concrete commits (cherry
> >> pick) before normally proceeding with the release process [2].
> >>
> >> So, what I just did this morning is to merge all changes we did in
> >> master into develop. This allows us to continue working on develop while
> >> we are in the voting process, and manage back in case it fails.
> >>
> >> Please, do not hesitate to ask any question about the workflow. I think
> >> the discussion would be quite rewarding for all of us.
> >>
> >> Kind regards,
> >>
> >> [1] http://marmotta.incubator.apache.org/development.html
> >> [2] http://wiki.apache.org/marmotta/ReleaseProcess
> >>
> >
> 



Re: branching workflow

Posted by Andy Seaborne <an...@apache.org>.
On 12/04/13 07:55, Sergio Fernández wrote:
> Any expert on git in the room? I thin we have a problem understanding
> the tagging system.
>
> According the documentation: when "git fetch" is run, unless told
> otherwise by an explicit --no-tags, it automatically fetches tags that
> point at new commits on branches you fetch.
>
> But this does not work for me, and I need to run "git fetch --tags" to
> actually retrieve the tags. Same for pushing.
>
> What this concept error is causing is: we removed some old tags, but not
> all clones did it, so on new push we are getting back those tags again.
>
> I'd really appreciate if someone could bring a proper explanation to
> such behaviour.
>
> Thanks!

(tries it out going RC6 to RC7 ...)

Do you mean specifically for "3.0.0-incubating"?

The reuse of the same name to move from RC(n) to RC(n+1) has given me 
some sync issues.

Yes - I needed "--tags".


git fetch
blah... blah

git checkout 3.0.0-incubating
HEAD is now at a3f52e9... [maven-release-plugin] prepare release 
3.0.0-incubating


git fetch --tags

 From https://git-wip-us.apache.org/repos/asf/incubator-marmotta
  - [tag update]      3.0.0-incubating -> 3.0.0-incubating

git checkout 3.0.0-incubating

Previous HEAD position was a3f52e9... [maven-release-plugin] prepare 
release 3.0.0-incubating
HEAD is now at 63762a3... [maven-release-plugin] prepare release 
3.0.0-incubating

	 Andy ("not a git expert")

>
> On 09/04/13 08:40, Sergio Fernández wrote:
>> Hi,
>>
>> this is something we already did in the previous release candidate, but
>> I think is interesting to describe with detail.
>>
>> According what we discusses back in January, we agreed on using a
>> branching workflow, as described at [1].
>>
>> Then, when the RC5 vote started, we opened a new develop branch where
>> continue working on all features would not go in the release, but for
>> 3.1.0-incubating-SNAPSHOT. We kept the master, and there I was fixing of
>> the issues with the release itself.
>>
>> Some of the changes pushed to develop where actually bug fixes that may
>> be interesting to release; so I merged those concrete commits (cherry
>> pick) before normally proceeding with the release process [2].
>>
>> So, what I just did this morning is to merge all changes we did in
>> master into develop. This allows us to continue working on develop while
>> we are in the voting process, and manage back in case it fails.
>>
>> Please, do not hesitate to ask any question about the workflow. I think
>> the discussion would be quite rewarding for all of us.
>>
>> Kind regards,
>>
>> [1] http://marmotta.incubator.apache.org/development.html
>> [2] http://wiki.apache.org/marmotta/ReleaseProcess
>>
>


Re: branching workflow

Posted by Sergio Fernández <se...@salzburgresearch.at>.
Any expert on git in the room? I thin we have a problem understanding 
the tagging system.

According the documentation: when "git fetch" is run, unless told 
otherwise by an explicit --no-tags, it automatically fetches tags that 
point at new commits on branches you fetch.

But this does not work for me, and I need to run "git fetch --tags" to 
actually retrieve the tags. Same for pushing.

What this concept error is causing is: we removed some old tags, but not 
all clones did it, so on new push we are getting back those tags again.

I'd really appreciate if someone could bring a proper explanation to 
such behaviour.

Thanks!

On 09/04/13 08:40, Sergio Fernández wrote:
> Hi,
>
> this is something we already did in the previous release candidate, but
> I think is interesting to describe with detail.
>
> According what we discusses back in January, we agreed on using a
> branching workflow, as described at [1].
>
> Then, when the RC5 vote started, we opened a new develop branch where
> continue working on all features would not go in the release, but for
> 3.1.0-incubating-SNAPSHOT. We kept the master, and there I was fixing of
> the issues with the release itself.
>
> Some of the changes pushed to develop where actually bug fixes that may
> be interesting to release; so I merged those concrete commits (cherry
> pick) before normally proceeding with the release process [2].
>
> So, what I just did this morning is to merge all changes we did in
> master into develop. This allows us to continue working on develop while
> we are in the voting process, and manage back in case it fails.
>
> Please, do not hesitate to ask any question about the workflow. I think
> the discussion would be quite rewarding for all of us.
>
> Kind regards,
>
> [1] http://marmotta.incubator.apache.org/development.html
> [2] http://wiki.apache.org/marmotta/ReleaseProcess
>

-- 
Sergio Fernández
Salzburg Research
+43 662 2288 318
Jakob-Haringer Strasse 5/II
A-5020 Salzburg (Austria)
http://www.salzburgresearch.at

Re: branching workflow

Posted by Sergio Fernández <wi...@apache.org>.
Another thing is I'd like to discuss to avoid the usage of "git pull". 
Basically this is the same than "git fetch && git merge",

I have read many articles explaining saying why 'git pull' is evil!, for 
example https://coderwall.com/p/jgn6-q , explaining the mess "git merge" 
could cause, recommending to use "git rebase" with origin.

I'd also appreciate opinions to respect on this :-)

On 09/04/13 08:40, Sergio Fernández wrote:
> Hi,
>
> this is something we already did in the previous release candidate, but
> I think is interesting to describe with detail.
>
> According what we discusses back in January, we agreed on using a
> branching workflow, as described at [1].
>
> Then, when the RC5 vote started, we opened a new develop branch where
> continue working on all features would not go in the release, but for
> 3.1.0-incubating-SNAPSHOT. We kept the master, and there I was fixing of
> the issues with the release itself.
>
> Some of the changes pushed to develop where actually bug fixes that may
> be interesting to release; so I merged those concrete commits (cherry
> pick) before normally proceeding with the release process [2].
>
> So, what I just did this morning is to merge all changes we did in
> master into develop. This allows us to continue working on develop while
> we are in the voting process, and manage back in case it fails.
>
> Please, do not hesitate to ask any question about the workflow. I think
> the discussion would be quite rewarding for all of us.
>
> Kind regards,
>
> [1] http://marmotta.incubator.apache.org/development.html
> [2] http://wiki.apache.org/marmotta/ReleaseProcess
>

-- 
Sergio Fernández