+1 for the validate button if there are cycles to make that happen or if it can reuse the existing feature (eg, if validating auto-sends to the project lead or specified email address).
But overall Im leaning even more in Johns direction that these do provide utility when running correctly. Perhaps not as much for Ed, but Ed is likely to be way more on top of this than others. On the Mylyn projects, I have repeatedly witnessed those updates trigger fixes. I dont think we should optimize reducing noise for projects that will never keep their metadata up-to-date at the cost of removing a useful trigger for those who want to, but are not quite on top of it.
Dr. Mik Kersten
Tasktop CEO, Mylyn Lead, http://twitter.com/mik_kersten
Assistant: zoe.jong@xxxxxxxxxxx, +1-778-588-6896, Skype: zoe.e.jong
I'm torn on this. It does have a fairly high noise-to-signal ratio but often it does point out valid omissions. Having recently done metadata setup for a new project, I definitely appreciated the validation that everything was set up correctly. As Ed suggests, having a button where I can validate the data myself without widely broadcasting the results would be handy. Maybe that, in combination with running it automatically once or twice a year would be enough.
I tend to regard keeping project metadata/websites/... up-to-date as one
of those "selfish-best interest" sorts of things. That is, projects
should keep this information up-to-date because it is good for the
health of the project.
Every month we have a script that runs to check that certain bits of
metadata are properly specified. Missing, or incorrect entries are
reported via email message to the project mailing list.
Every month, a couple of project leads contact me and the Webmaster
about the messages, generally because something that's being reported
doesn't make sense (e.g. reports that a next release isn't specified
when one seems to have been specified). Every month, the Webmasters and
I spend a good couple of hours chasing down the problem and tend to end
up modifying the check scripts. These good couple of hours could be
better spent doing other more valuable things for the committers and
Most projects--it seems--just ignore these messages anyway.
So... I've been thinking. Is it time to just turn off this notification
mechanism? Or maybe just run it twice a year or something?
Your thoughts are appreciated.
The Eclipse Foundation
eclipse.org-architecture-council mailing list
IMPORTANT: Membership in this list is generated by processes internal to the Eclipse Foundation. To be permanently removed from this list, you must contact emo@xxxxxxxxxxx to request removal.