Re: svn commit: r765901 - /incubator/river/jtsk/skunk/jcosters/

[Jukka Zitting <ju...@gmail.com>]

Hi,

On Fri, Apr 17, 2009 at 10:41 AM,  <jc...@apache.org> wrote:
> Create personal sandbox branch of jtsk/trunk.

IMHO it would be better if we didn't have personal sandboxes. If you
have experimental changes that shouldn't yet go to trunk, it would be
better to put it in a topic-based branch that clearly identifies the
purpose of that experiment. Having personal branches discourages
cooperation between project members.

The same applies to the earlier branch created by Niclas.

BR,

Jukka Zitting

Re: svn commit: r765901 - /incubator/river/jtsk/skunk/jcosters/

[Jonathan Costers <jo...@googlemail.com>]

Op vrijdag 17-04-2009 om 14:11 uur [tijdzone +0200], schreef Jukka
Zitting:
> Hi,
> 
> On Fri, Apr 17, 2009 at 2:06 PM, Jukka Zitting <ju...@gmail.com> wrote:
> > IMHO it would be better if we didn't have personal sandboxes. If you
> > have experimental changes that shouldn't yet go to trunk, it would be
> > better to put it in a topic-based branch that clearly identifies the
> > purpose of that experiment.
> 
> Looking at the changes you've committed, I think they could have gone
> directly to trunk. Is there a reason why you'd rather use a branch for
> those changes?
> 
> BR,
> 
> Jukka Zitting

I am doing some last tests on two development machines and committing
changes to my skunk branch as they come along. It's easy for me to keep
my dev boxes synced that way.
As soon as this is finished, and nobody has objections, I can merge my
skunk branch to the trunk and get rid of it :-)

Best
Jonathan

Re: svn commit: r765901 - /incubator/river/jtsk/skunk/jcosters/

[Jukka Zitting <ju...@gmail.com>]

Hi,

On Fri, Apr 17, 2009 at 2:06 PM, Jukka Zitting <ju...@gmail.com> wrote:
> IMHO it would be better if we didn't have personal sandboxes. If you
> have experimental changes that shouldn't yet go to trunk, it would be
> better to put it in a topic-based branch that clearly identifies the
> purpose of that experiment.

Looking at the changes you've committed, I think they could have gone
directly to trunk. Is there a reason why you'd rather use a branch for
those changes?

BR,

Jukka Zitting

Re: svn commit: r765901 - /incubator/river/jtsk/skunk/jcosters/

[Jonathan Costers <jo...@googlemail.com>]

I'll merge my stuff to the trunk in a couple of hours.

Thanks for your advice!

Best
Jonathan

Op vrijdag 17-04-2009 om 14:32 uur [tijdzone +0200], schreef Jukka
Zitting:
> Hi,
> 
> On Fri, Apr 17, 2009 at 2:19 PM, Jonathan Costers
> <jo...@googlemail.com> wrote:
> > My intention was to let people check out my changes for RIVER-301
> > together with changes for RIVER-272, in an easy way. The patch I
> > attached to RIVER-301 was a bit too big to handle.
> 
> OK, cool. For such cases I propose that the branch is named after the
> issue in question, so this could have been skunk/RIVER-301 or
> skunk/RIVER-301-with-272.
> 
> > In my opinion, we can commit these changes to the trunk too, but I
> > wanted to wait for reaction first.
> 
> Personally I'd just commit directly to trunk as long as you're not
> explicitly breaking things for others. It's easy to revert changes in
> case problems are identified.
> 
> PS. There was quite a bit of earlier discussion about having other
> committers review your changes before committing them. It's a good
> idea in general especially for more complex changes, but currently we
> have so few active committers, that I'd rather opt for the occasional
> mistakes than for halting development due to insufficient reviews. As
> an example see Mark's recent commit of a patch that waited six months
> in the issue tracker with nobody to review it.
> 
> BR,
> 
> Jukka Zitting

Re: svn commit: r765901 - /incubator/river/jtsk/skunk/jcosters/

[Jukka Zitting <ju...@gmail.com>]

Hi,

On Fri, Apr 17, 2009 at 2:19 PM, Jonathan Costers
<jo...@googlemail.com> wrote:
> My intention was to let people check out my changes for RIVER-301
> together with changes for RIVER-272, in an easy way. The patch I
> attached to RIVER-301 was a bit too big to handle.

OK, cool. For such cases I propose that the branch is named after the
issue in question, so this could have been skunk/RIVER-301 or
skunk/RIVER-301-with-272.

> In my opinion, we can commit these changes to the trunk too, but I
> wanted to wait for reaction first.

Personally I'd just commit directly to trunk as long as you're not
explicitly breaking things for others. It's easy to revert changes in
case problems are identified.

PS. There was quite a bit of earlier discussion about having other
committers review your changes before committing them. It's a good
idea in general especially for more complex changes, but currently we
have so few active committers, that I'd rather opt for the occasional
mistakes than for halting development due to insufficient reviews. As
an example see Mark's recent commit of a patch that waited six months
in the issue tracker with nobody to review it.

BR,

Jukka Zitting

Re: svn commit: r765901 - /incubator/river/jtsk/skunk/jcosters/

[Jonathan Costers <jo...@googlemail.com>]

Op vrijdag 17-04-2009 om 14:06 uur [tijdzone +0200], schreef Jukka
Zitting:
> Hi,
> 
> On Fri, Apr 17, 2009 at 10:41 AM,  <jc...@apache.org> wrote:
> > Create personal sandbox branch of jtsk/trunk.
> 
> IMHO it would be better if we didn't have personal sandboxes. If you
> have experimental changes that shouldn't yet go to trunk, it would be
> better to put it in a topic-based branch that clearly identifies the
> purpose of that experiment. Having personal branches discourages
> cooperation between project members.
> 
> The same applies to the earlier branch created by Niclas.
> 
> BR,
> 
> Jukka Zitting

I agree.
My intention was to let people check out my changes for RIVER-301
together with changes for RIVER-272, in an easy way. The patch I
attached to RIVER-301 was a bit too big to handle.

In my opinion, we can commit these changes to the trunk too, but I
wanted to wait for reaction first.

Best
Jonathan