Re: [ecf-dev] feature refactoring

[Scott Lewis <slewis@xxxxxxxxxxxxx>]

On 1/29/2014 3:06 PM, Wim Jongman wrote:

;) i meant to say "Tycho nor Buckminster" But I hear ant is popular again these days ...

That's wonderful.  :)

All goofing aside: If we refactor all features then also the complete build must be refactored. Maybe we should focus on the core features from branch bugxxxxx first and leave the rest

Yeah...as of today, the core and filetransfer features are pretty much done on bug409787 branch.   There will likely be build changes necessary, but I don't think they will be huge.

so that we don't have to stress to make the release train.

Well, the main thing about support of p2/Platform is that we need to get some of David Williams' time (and perhaps p2 folks' time)... to complete and test the releng changes on the platform side (for them to consume ECF filetransfer features directly rather than consuming the bundles in p2/platform features as they have been doing previously).   The changeover also has to happen on their Luna milestone schedule [1], which probably means targeting the M6 milestone (probably 1 week prior to March 14) for having everything changed over, tested, working in Platform build, etc.

But I'm hopeful that we can get a build of these now refactored features pretty easily, and provide it to David sometime in the next few weeks.  I don't particularly care what build system we use to do it, but it seems to me that it's unlikely to happen with Tycho in the next few weeks (although I would be happy to have someone prove me wrong).

Scott

[1] http://wiki.eclipse.org/Luna/Simultaneous_Release_Plan#Schedule

On Wed, Jan 29, 2014 at 11:47 PM, Scott Lewis <slewis@xxxxxxxxxxxxx> wrote:

On 1/29/2014 2:29 PM, Wim Jongman wrote:

What is important is the build of these new features for platform consumption, whether we build with Tycho or Bucky.

I would amend this a little...what's most important to me is the build of all the ECF features...for ECF consumer consumption.   I'm hopeful that these feature refactoring changes will make our interaction with the platform consumption easier...both on us, them, and on our consumers (which is my major concern wrt 409787).

For most people, getting a trivial build going with Tycho is much easier than it is with Bucky.

Perhaps true, but

a) this is not a trivial build...it's go multiple pieces/features, which dependencies between them and other other things, signing, automated testing, automated running of OSGi CT tests, etc.
b) we already have both the build and the expertise with our current Bucky-based build
c) I'm not personally able to make the commitment to learning, using, and redoing all that we already have going with Tycho...in the short term at least
d) we are going to need some additional/new releng efforts over the next few years...but this is not unusual for us...it's also true for other new parts of ECF (new providers, new APIs, etc).

Unless we are going to rely on M. for the next couple of years, your only choice is switching to Tycho or hope for Buckminster to awake from the dead.

Well, it's my hope that someone new will step up.   Markus (and to a lesser extent me and you) has been doing wonderful service on ECF releng for the past few years.

> (i.e. someone is going to have to commit to maintain the Tycho build for us over the next few years).

I like your recruiting method but I don't think you find someone to fill this profile for Tycho or Buckminster.  ;-)

Why do you say that?  Is the releng/tools world going back to Makefile?  ;-)

Scott

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

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