Should maint branch be deprecated?

Should maint branch be deprecated?

[Martyn Jago]

As an infrequent committer to org-mode, I wonder if the maint branch is
more pain than gain. 

I've read through the mails arguing for its conception, and the benefits
just don't appear to be resultant.

Add to that - infrequent committer's who commit via the list generally
know nothing about the maintenance branch (including me), which just
leads to more confusion.

I follow the commits to master, and the work to core developers due to
the maint branch is clearly not insignificant?

If the gain is greater than the pain then somebody needs to clue me up.

Best, Martyn

Re: Should maint branch be deprecated?

[Bastien]

Hi Martyn

Martyn Jago <martyn.jago@btinternet.com> writes:

> As an infrequent committer to org-mode, I wonder if the maint branch is
> more pain than gain. 

I'm still in the process of fixing things.  

After I'm done, I will suggest a new workflow.

Thanks,

-- 
 Bastien

Re: Should maint branch be deprecated?

[Achim Gratz]

Martyn Jago <martyn.jago@btinternet.com> writes:
> As an infrequent committer to org-mode, I wonder if the maint branch is
> more pain than gain. 
>
> I've read through the mails arguing for its conception, and the benefits
> just don't appear to be resultant.

It seemed to work well until Bastien made a mistake because the release
script wasn't properly adapted and digged the hole deeper while trying
to fix it (too) quickly.  Maint is good again, master still needs a few
more touches.

> Add to that - infrequent committer's who commit via the list generally
> know nothing about the maintenance branch (including me), which just
> leads to more confusion.
>
> I follow the commits to master, and the work to core developers due to
> the maint branch is clearly not insignificant?

It is actually quite easy:

If it's a bugfix for something broken in a release version, commit to
maint and merge maint back into master.

If implementing a new feature or fixing something not yet released,
commit to master.

> If the gain is greater than the pain then somebody needs to clue me up.

The argument is the same as before: master contains changes that
introduce new features and may break compatibility while maint should
only contain bugfixes.  This is important because org also needs to
respect Emacs' release schedule (and synchronize with the bzr repo they
use), which generally means that they will accept bugfixes, but no new
features during certain times (like right now).  The other development
model is to develop new stuff in feature branches and those tend not to
get tested too well before going mainline.

Ultimately it's Bastiens' call.  If he decides to go back to the single
branch model, he'll have to do much more tedious manual work to keep org
in sync with Emacs and when doing bugfix releases.


Regards,
Achim.
-- 
+<[Q+ Matrix-12 WAVE#46+305 Neuron microQkb Andromeda XTk Blofeld]>+

Wavetables for the Terratec KOMPLEXER:
http://Synth.Stromeko.net/Downloads.html#KomplexerWaves

Re: Should maint branch be deprecated?

[Bastien]

Hi Achim,

Achim Gratz <Stromeko@nexgo.de> writes:

> It seemed to work well until Bastien made a mistake because the release
> script wasn't properly adapted and digged the hole deeper while trying
> to fix it (too) quickly.  Maint is good again, master still needs a few
> more touches.

This should be all right now -- thanks again!

>> If the gain is greater than the pain then somebody needs to clue me up.
>
> The argument is the same as before: master contains changes that
> introduce new features and may break compatibility while maint should
> only contain bugfixes.  This is important because org also needs to
> respect Emacs' release schedule (and synchronize with the bzr repo they
> use), which generally means that they will accept bugfixes, but no new
> features during certain times (like right now).  

See my email about the new git workflow: I want to have a branch
dedicated to releases.  I think it will make things way more readable
and manageable.

> The other development
> model is to develop new stuff in feature branches and those tend not to
> get tested too well before going mainline.

We already use dedicated branch for development sometimes -- see the
branch for Max/Carsten new feature.  But you're right in saying that 
I want to keep the majority of development happening in master.

> Ultimately it's Bastiens' call.  If he decides to go back to the single
> branch model, he'll have to do much more tedious manual work to keep org
> in sync with Emacs and when doing bugfix releases.

With the old model, the *only* reason for having maint was the Emacs
sync issue.  Now the reason is to clarify the release process.

Best,

-- 
 Bastien

Re: Should maint branch be deprecated?

[Bastien]

Hi Martyn,

Martyn Jago <martyn.jago@btinternet.com> writes:

> As an infrequent committer to org-mode, I wonder if the maint branch is
> more pain than gain. 

It is a pain when too many things should happen on it.  

IMO It is a gain when we dedicate it to releases only.

No pain should come out of introducing hotfix-* branches, 
as they are transient branches.

Now I'm off to a well-deserved bed :)

-- 
 Bastien

Re: Should maint branch be deprecated?

[Martyn Jago]

Bastien <bzg@altern.org> writes:

Hi Bastien, Achim

> Hi Martyn,
>
> Martyn Jago <martyn.jago@btinternet.com> writes:
>
>> As an infrequent committer to org-mode, I wonder if the maint branch is
>> more pain than gain. 
>
> It is a pain when too many things should happen on it.  
>
> IMO It is a gain when we dedicate it to releases only.
>
> No pain should come out of introducing hotfix-* branches, 
> as they are transient branches.
>
> Now I'm off to a well-deserved bed :)

I certainly appreciate the need to stabilize releases. But I did get to
wondering lately if the original purpose of maint was kind of lost in
the overall high-throughput repo activity. 

I'm a big fan of Git, and have started using it for non-version control
activities, so I appreciate things can always be fine-tuned to suit the
current situation.

Thanks for the considered replies.

Best, Martyn