[wtp-releng] Two changes coming up in WTP builds.

[David M Williams <david_williams@xxxxxxxxxx>]

For your awareness ...

325039
- Should sign all builds and bundles 
325181
- We should use "comparator" to detect changed bundles (and deliver
only changed ones)

The first one I mentioned in previous
messages .. but have been meaning to call your attention to it. We'll start
this immediately for I-builds, and any subsequent M-builds (we always sign
M builds, but not, so far, the unit test bundles). We've always signed
S builds and R builds ... but not always I-builds, trying to save time
and processing. It'll be better to sign them all, and find better ways
to shorten the build (e.g. publishing early, before tests are ran, and
stopping if compile errors found, etc). 
 
I think the only downside of signing
all bundles, every build, is that it will take longer to get a complete
build. In theory, there are a few cases where signing might break some
code, so let us know if you think you've found one of those cases in your
tests. There is also the chance it can slow things down running test code
... but ... our tests take so long to complete, I doubt we'd notice a signing
slow down (ha ha). 

The "comparator" issue might
end up effecting you directly. Its pretty cool, but I can imagine it'll
take some time to get used to it and could lead to "mysterious"
errors. Especially until we have some good tests in place to detect "real"
problems. 
In the end, once fully implemented,
I could imagine sometimes there will be cases where a committer might end
up thinking "I know I changed that constant in the commons module"
... "why isn't that constant being picked up in the JEE module".
In fact, it might have been picked up during the build, but if the version
didn't change in JEE module, we would "throw away" the new version
and use the old version.  There might be times you will have to tag
dependent source bundles, if there are these type of "build time only"
effects that changes your byte codes. 

I mention these now in case you see
changes in builds or find mysterious results. 

I think this will end up being goodness
... but, we all may have to educate ourselves a little more on OSGi versioning
and p2 behavior ... I know, your favorite topics . :) 

Plus, there's been some more minor changes
going on which lead to issues, for example, the "wst zip" is
not being produced at the moment. I hope to get that all cleaned up and
restored some time this week. Let me know if you see/find other things
that get in your way.