Skip to main content

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index] [List Home]
Re: [cross-project-issues-dev] Juno M6 is not far away ... are you still greedy?

Hi David,

Thank you for pointing our all these issues. Unfortunately this confuses me a lot. Some questions:

1) Do I have to understand all this? Or
2) Is it just a matter of using a 3.8 Buckminster? And if so
2.1) Where is it?

Cheers
/Eike

----
http://www.esc-net.de
http://thegordian.blogspot.com
http://twitter.com/eikestepper



Am 06.03.2012 04:39, schrieb David M Williams:


I've added a new report to the common repository reports [1].

It is called "Optional Runtime Requirements and Greediness" [2]

I doubt it is too useful at the moment, except to reflect the magnitude of
the problem ... that many projects still have to move up to a more recent
publisher. Ideally, by M6 or M7, the first 2 or 3 lists in that report will
all be "0" (indicating we all are using the new p2 publisher) but they are
currently in the hundreds. The reports do not currently track back to "who
is responsible" for using the old publisher, but I might add that when the
list gets smaller. As it is, the reporting code is simply looking at only
the "<required" elements to get the listings, and not keep track of which
IU required it.

As I mentioned in another note on this topic[3], the p2 publisher as of
Juno M1 began to publish "greedy=false" for those requirements which are
optional at runtime. The previous, implied default was "greedy=true".  This
better meets users and adopters expectations.

According to Thomas Hallgren[1], the Buckminster builder will use the new
publisher "by M6" so, not sure if that will be in time for you
Buckminster-builders to take advantage of for your own M6 contribution to
Juno M6, but you might keep a close eye on it, and plan on changing shortly
after M6 if not before.

According to Igor Fedorenko [5] Tycho uses 3.8-based publisher and
generates non-greedy optional dependencies since version 0.13. So, I hope
everyone who uses Tycho has or can soon move up to that version (or, what
ever is their latest recommended version).

So, if you haven't yet, please do try and move up to a Juno based p2
publisher as soon as possible so we can do more testing on what we will
release and find problems or surprises early. I suspect some of the
anomalies in the report (such as com.ibm.icu sometimes required greedily
and sometime not) is a side effect of people using "old" Orbit builds for
their deliverables.? Orbit is planning to provide a Juno M6 usable S-build
this Friday 3/9, so everyone should be able to move up to latest Orbit S
build next week.

While the reports may not be too useful to your individual project at the
moment (since, it is just a big list, without saying who's to "blame") you
might find it interesting. I thought it would make a good excuse to give a
timely reminder to upgrade your p2 publisher.

Remember, M6+0 is 03/16 and M6 complete and available on Friday, 03/23 [6]
which is significant because that is immediately before EclipseCon! It is a
challenge every year to make sure we get it done before people have to
start traveling!

Much thanks,


[1] http://build.eclipse.org/juno/simrel/
[2]
http://build.eclipse.org/juno/simrel/reporeports/reports/greedyReport.html
[3]
http://dev.eclipse.org/mhonarc/lists/cross-project-issues-dev/msg07098.html
[4] http://dev.eclipse.org/mhonarc/lists/buckminster-dev/msg01320.html
[5]
http://dev.eclipse.org/mhonarc/lists/cross-project-issues-dev/msg07100.html
[6] http://wiki.eclipse.org/Juno/Simultaneous_Release_Plan



Back to the top