I moved the targets of these bugs to 3.4.2, they should get sufficient testing in 3.5M2+ for us to be more comfortable putting things in for 3.4.2
As a request, I would like to see 3.4.2 builds earlier so people can pick them up if necessary.
Thanks all.
On Tue, Sep 9, 2008 at 2:30 AM, Philippe P Mulet
<philippe_mulet@xxxxxxxxxx> wrote:
I would share Tom's concern that we
have no decent regression tests for this new scenario.
Would 3.4.2 be a more better target,
buying us time to challenge the new behavior and construct regression test
suites ?
For urgent needs, we could always construct
a patch feature as an early preview of a 3.4.2 fix.
We are in 3.4.1 rc3 stage, and having
this request show up only now feels a bit wrong; typically I would have
expected such a change to go in right after 3.4.0 so it could be exercized.
Right. The key really is the risk that the bits
previously generated by PDE UI are different somehow than the bits newly
generated by PDE build. The Build code has been around and used for
a long time but this is a new scenario. Personally I would very much
like to see this bug fixed and do believe it is an issue that it is a critical
bug that real people will encounter. Using headless build as an alternative/workaround
certainly works but it is really quite a barrier to adoption and use.
Having said that it is late in the cycle for 3.4.1 and we don't have much
time to test on different platforms etc. The best laid plans and
all that... If we are going to do this we need several people to
look at this, test it and give their ok.
Jeff
Chris Aniszczyk wrote:
In terms of tests, unfortunately it's all manual for exports,
not automated.
In the end, this has to go in 3.4.X otherwise people who use p2 and export
from the UI are broken. The workaround is to use a headless build. However,
there has been a few bugs coming in already with issues in regarding to
product export using UI so this is why I have urgency to put this in. This
basically eliminates any difference between PDE UI and PDE Build when generating
products.
~50 LOC are being removed in PDE UI that now delegate to PDE Build
1 LOC being changed in PDE Build to remove the PDE UI killswitch from using
product-based export
On Mon, Sep 8, 2008 at 3:40 PM, Thomas Watson <tjwatson@xxxxxxxxxx>
wrote:
Maybe it is just me, but this sounds like a risky change
for a point release. Especially after we have entered the RC stage of 3.4.1.
Chris, is this really low risk? How many tests do we have to ensure that
no regressions will be introduced? Just curious, how many lines of code
are being removed and are we sure the PDE-Build behavior gives us all the
behavior we expect in this scenario?
Tom
"Chris
Aniszczyk" ---09/08/2008 03:23:12 PM---(pde ui) product export from
the ui does not generate bundles.info
Sauf indication contraire ci-dessus:/ Unless stated otherwise above:
Compagnie IBM France
Siège Social : Tour Descartes, 2, avenue Gambetta, La Défense 5, 92400
Courbevoie
RCS Nanterre 552 118 465
Forme Sociale : S.A.S.
Capital Social : 542.737.118 euros
SIREN/SIRET : 552 118 465 02430
_______________________________________________
eclipse-pmc mailing list
eclipse-pmc@xxxxxxxxxxx
https://dev.eclipse.org/mailman/listinfo/eclipse-pmc
--
Cheers,
~ Chris Aniszczyk
