Stephen Lau wrote:
John Levon wrote:
This commit:
http://mail.opensolaris.org/pipermail/onnv-notify/2007-January/006656.html
has 20 changesets since the dboot putback broke the bridge. stevel
tells me
it's intentional that the commit comment doesn't list all the comment
contents
of the changesets; why?
regards
john
If I recall the conversation correctly, it was to prevent the message
from getting astronomically large if someone pulled-a-stevel and
imported all of onnv-gate (which would generate a notification
containing every single putback comment that had gone into onnv-gate).
However we've slapped a check in to automatically reject anything with
>500 incoming changesets - so perhaps we can modify the behaviour to
show all the comment contents of the changesets.
I'd prefer that since the notification would be more useful then,
especially given that we aggregate the file listing. As you noted on
IRC, that putback notification makes it look like a relatively minor FMA
putback updated a helluva lot of files. :)
I can't easily think of a pleasant way to deal with this, there's many times
when either side is the wrong thing to be doing (condensing or otherwise).
I'd say that the confusion with the filelist is wrong either way, however.
(except with merges, where it's actually somewhat accurate...)
I still think the better solution is to make the accidental mailbombing
problem go away by providing better setup facilities, and then consider what
we need to do for the intentionally large merge side of things.
My current hopes would be merges only show the merge cset (which maybe
complicated to implement in a foolproof fashion), others send one
mail-per-change, and putbacks of large numbers of changesets are blocked on
the assumption they're bogus, and dealt with case-by-case for the situation
where they were intended (but still perhaps unwise).
-- Rich
_______________________________________________
tools-discuss mailing list
tools-discuss@opensolaris.org
