[wtp-dev] Why do we need Galileo?

[David M Williams <david_williams@xxxxxxxxxx>]

I'm sure there's a number of good reasons,
but what are they? I'd like to ask we in WTP do a better job of understanding
what we specifically need or use in Galileo. 

Put another way, we are (finally) getting
pretty good about setting the upper bounds of pre-req ranges correctly,
but I'm not sure we give much thought to the lower bounds, and maybe it's
about time we thought about it ... maybe in future releases we might have
specific goals or plans, but for now, I'm just trying to understand. 

This is actually a hotly debated and
very controversial issue in the Eclipse community. Some argue the lower
bound should be set as high as "the current pre-req stream" since
that's what is tested (and some even do that automatically in their builds!).
Some argue it should be as low as possible and only increase lower bound
plugin by plugin when it's concretely known to be required for some specific
reason. From what I know, even what is considered "required"
isn't always so clear and may differ from project to project. 

I'm not trying to settle that controversy.
I'm just trying to understand specific cases in WTP better ... both for
myself, and I think we as a Project are getting mature enough we should
be able to state "why we need Galileo" as a pre-req with some
degree of precision.  I see this as similar to the way we say we need
a certain level of JRE or even a certain pre-req in general (such as "why
do we need EMF" :) . 

I've written a wiki page that lists
some of the high level issues, and started a (very small) list of specific
cases in WTP which I just happen to know about:
http://wiki.eclipse.org/WTP/Architecture/A_study_of_prereq_lower_bounds

I'd like to ask all committers to participate
and update the document whenever you find a reason to update the lower
bound of one of your pre-reqs. Just briefly, no long explanations required,
but reference to a bug number would be handy. I'm thinking if we can document
10 to 20 or more cases that it might start to get interesting. I'd expect
most items to fall into a few high-level categories ... make use of a whole
new function, some improved API or to pick up required bug fixes, being
obvious ones ... but I also wonder if there are many cases where behavior
changes in subtle ways, requiring us to code differently for Galileo, or
... what else? How many? That's what I don't know and would like your help
to educate me (and each other).  While hypothetically we might have
tools to list out which plugins pre-req the latest version of Galileo,
I don't think that would be all that useful ... the first question then
would still be "why" so I'm hoping we can just get directly to
"why".  As always, suggestions welcome. 

We can discuss more at status meetings,
I just wanted to send out this note and wiki link in advance. 

Thanks,