Skip to main content

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index] [List Home]
Re: [tycho-dev] tycho p2 fork

p2 fork for Tycho is an unfortunate necessity. The good news is, the
patches in question have been pushed to the main p2 repository already,
so we'll switch back to official p2 builds in a matter of days.

--
Regards,
Igor

On 11-12-12 10:52 AM, Oberlies, Tobias wrote:
Just too summarize my point of view after this discussion:

- I agree that we must not release Tycho with a p2 fork.
- I don't think that creating a p2 fork for Tycho is a good idea: It just delays the point where the p2 patches have to be integrated, and that that point in time the author of the patch may not be able to do the necessary work, e.g. additional tests, to get the patch in. Having the branch encourages to build up technical debt, and this is obviously a bad idea.
- I will review p2 patches of Tycho contributors as part of the (limited) work I do for Tycho, but I will not automatically take the responsibility for integrating all changes in the p2 fork into p2 master.

Best regards
Tobias


-----Original Message-----
From: Sievers, Jan
Sent: Donnerstag, 8. Dezember 2011 08:38
To: Igor Fedorenko
Cc: Oberlies, Tobias; jvanzyl
Subject: RE: tycho p2 fork

sounds reasonable.
just need to make sure we don't leak the P2 fork into the tycho release.

If we can get patches into p2 more quickly than in the past, consuming p2
nightly builds during tycho SNAPSHOT development could be a similar
option.

@Tobias what do you think?

Jan

-----Original Message-----
From: Igor Fedorenko [mailto:ifedorenko@xxxxxxxxxxxx]
Sent: Mittwoch, 7. Dezember 2011 20:15
To: Sievers, Jan
Cc: Oberlies, Tobias; jvanzyl
Subject: Re: tycho p2 fork

Answering to myself, but I think I have a workable middle-ground solution

What if we use P2 fork only for SNAPSHOT Tycho builds and only when
there is a delta between Tycho P2 fork and main P2 source tree? And use
official P2 builds for Tycho RELEASE builds and whenever there are no
outstanding P2 changes.

This will allow all Tycho developers make necessary P2 changes and
include them in Tycho SNAPSHOT builds, which addresses my main concern.
It will also allow Tobias switch Tycho builds back to the official P2
builds once he's reviewed and merged the P2 changes, which I believe
addresses your concerns.

--
Regards,
Igor



On 11-12-07 12:48 PM, Igor Fedorenko wrote:
You and I and any new Tycho committer to join the project are
essentially locked out of large part of the possible work we can do in
Tycho. This significantly impedes Tycho development and it is my
responsibility as the Tycho project lead to remove this impediment. P2
fork is one solution this problem but I am certainly open to discuss
alternatives.

--
Regards,
Igor

On 11-12-07 11:06 AM, Sievers, Jan wrote:
Tobias is a p2 committer. I doubt that he reviewed your patches.
If they are important for tycho they should also be important for him.

Did he commit to do anything and then did not follow up on it in a
timely manner?
If that's the case I can understand you are upset but let's get back
to the technical problems.

What are the bugzilla numbers we are talking about here?

Regards,
Jan

-----Original Message-----
From: Igor Fedorenko [mailto:ifedorenko@xxxxxxxxxxxx]
Sent: Mittwoch, 7. Dezember 2011 16:42
To: Sievers, Jan
Cc: Pascal Rapicault; jvanzyl; Oberlies, Tobias
Subject: Re: tycho p2 fork

I am honestly not interested in a temporary fix any more. I don't want
to beg anybody to consider Tycho-reladed P2 changes every time I need
to
fix a bug or implement a new feature in Tycho. P2 is at the core of
Tycho functionality and being able to make changes in P2 is essential
to
continue Tycho development.

Do you have a solution that addresses this fundamental problem?

--
Regards,
Igor

On 11-12-07 10:29 AM, Sievers, Jan wrote:
Maybe it's easier to sort this out on the phone.

Tobias is sick today but as soon as he returns to the office, I will
let you know and we can schedule our usual conf call.

For the call, I propose to get back to the technical reasons why you
think a fork is the only choice we have.

- What exactly are the most pressing bugs/missing features in p2
right now from your point of view
- Why is the patch not (yet) accepted, did Tobias review the patch
yet, etc.

Regards,
Jan

-----Original Message-----
From: Igor Fedorenko [mailto:ifedorenko@xxxxxxxxxxxx]
Sent: Mittwoch, 7. Dezember 2011 15:49
To: Sievers, Jan
Cc: Pascal Rapicault; jvanzyl; Oberlies, Tobias
Subject: Re: tycho p2 fork

git makes both tracking local changes and merging with upstream
incredibly easy. The choice we have is between the cost of maintaining
the fork and essentially stalling Tycho development. The fork was too
expansive with git and is quite affordable with git, especially when
compared with other choices we have.

--
Regards,
Igor

On 11-12-07 8:51 AM, Sievers, Jan wrote:
I've had the pleasure to clean up some of the forks we did in the
past (classes copied from maven-surefire, org.eclipse.osgi) and I can
say it's not a pleasure. I had to do a substantial amount of software
archaeology and debugging to find out why the fork was done in the
first place.

I have some doubts here because we don't have a particular track
record on following up on patches in other projects and making sure
they are accepted upstream.

This is technical debt. Sure it's cheap to fork now but what about
the long-term effects?

Regards, Jan>
-----Original Message-----
From: Pascal Rapicault [mailto:pascal@xxxxxxxxxxxx]
Sent: Mittwoch, 7. Dezember 2011 14:41
To: Sievers, Jan
Cc: Igor Fedorenko; jvanzyl; Oberlies, Tobias
Subject: Re: tycho p2 fork

I'm honestly not bothered with the fork. Given your needs to execute
promptly on your deliverables this is probably the best solution.

On 2011-12-07, at 3:05 AM, Sievers, Jan wrote:

We have already made some efforts so that now we consume recent
milestones of p2 in tycho.
We have a p2 committer on the tycho project.

My expectation is that we should be able to sort this out without
forking.

Regards
Jan

-----Original Message-----
From: Pascal Rapicault [mailto:pascal@xxxxxxxxxxxx]
Sent: Mittwoch, 7. Dezember 2011 02:21
To: Igor Fedorenko
Cc: Wayne Beaton; Andrew Ross; jvanzyl; Sievers, Jan; Oberlies,
Tobias
Subject: Re: tycho p2 fork

Thanks for the heads up.
Let's just make sure that this is a temporary situation (a couple
months) and that we can find a way to integrate all the necessary
changes back into p2.


On 2011-12-06, at 4:32 PM, Igor Fedorenko wrote:

Hello,

I am working under contract with Eclipse Foundation to setup Tycho
build
for the Eclipse Platform. This work is currently blocked by
outstanding
P2 issues. In order to continue to make progress with the Eclipse
Platform Tycho build work I will create Tycho-specific P2 fork soon
after Juno M4 is declared. Starting with version 0.14 Tycho will
ship
with modified P2 binaries.

Please note that this is NOT an attack on P2 develops, but merely
acknowledgement of the fact that the two projects have different
priorities and development schedules. For example, excessive
amount of
remote repository requests is one of the most complained about
Tycho
problems. At the same time, the problem does not affect other P2
users
to the same extend and the underlying P2 limitation [1]
understandably
is not treated as high priority issue by P2 developers.

I believe ability to develop and deliver Tycho-specific P2
enhancements
is in the best interested of Tycho user community and it will not
prevent P2 developers to absorb Tycho-related changes according to
their
development priorities and schedule.

@Wayne from IP standpoint, do I need to do any extra paperwork to
track
changes I make on the fork or I can treat it as part of Tycho
source tree?


[1] https://bugs.eclipse.org/bugs/show_bug.cgi?id=337022


--
Regards,
Igor


_______________________________________________
tycho-dev mailing list
tycho-dev@xxxxxxxxxxx
https://dev.eclipse.org/mailman/listinfo/tycho-dev


Back to the top