To provide a contrary story, we at TI have
always found that we can never take Eclipse or CDT “as-is”. We
hate doing it, and we’re actively trying to help out so that this delta
hopefully shrinks to nothing, but I don’t know if this will ever truly
happen since we rather push the envelope and use CDT as a building block and
not just a total solution in and of itself.
There will always be some bug discovered by
our internal QA that some project manager says we must fix for our product
release, or some feature that our customers say they must have, etc., and it’s
very difficult to just wait for the next CDT release for what you need and
pitch your own product schedule out the window.
___________________________________________
Chris Recoskie
Texas Instruments, Toronto
From:
cdt-dev-bounces@xxxxxxxxxxx [mailto:cdt-dev-bounces@xxxxxxxxxxx] On Behalf Of Lott, Jeremiah
Sent: Friday, July 22, 2005 10:29
AM
To: CDT General developers list.
Subject: RE: [cdt-dev] CDT 3.0
Closing
David Daoust:
> There is a wider community that
takes our published dates and makes plans around them. These plans
may
> include just taking the CDT
unchanged, or they may include scheduling a snapshot + fixes + testing.
Either
> way, when we are sloppy on our
dates, we negatively affect others.
Alain Magloire:
> This may sound harsher then intended, but companies
shipping major piece of code, GCC, GDB, Mozilla
> etc
… will usually take a snapshot/release and do there own Q/A before
incorporating the open source
> project.
Eclipse, CDT are no exceptions.
Just a quick note on our part.
I don't know what others do, but it has become our preference to ship CDT
(and Eclipse) with no additional patches. This ensures maximum
compatibility between our product and plugins from other vendors. This
has generally worked pretty well because both CDT and Eclipse follow a release
cycle that is more like a corporate setting. Releases are (compared to
other open source projects) infrequent and undergo QA by the community before
being declare a "release". Dates are set and we can plan a
release, knowing that the actual release date will be pretty close. If
major defects are found, the community reacts to quickly make a new bugfix
release. We obviously do our own QA, but we do somewhat expect that
before CDT was declared "released" that it underwent a decent QA
process, which has held pretty true. So I guess we do treat CDT and Eclipse
as exceptions to other open source projects.
Separately to specifically address the
issue at hand. For us I think priority would be 1. production quality, 2.
on-time release, 3. features. So I would prefer that minor annoyances or
issues with workarounds, as well as unimplemented features, would be pushed off
to ensure a more effective QA cycle. Of course I'm not a committer, so I
guess that's my $.01 :).
|