Re: gerrit lame / non-reviews spammage

Hi,

On Tue, Jul 10, 2012 at 11:28:01AM +0100, Michael Meeks wrote:
>       Hard to see of course; then again - if we could substantially shrink
> the content of the mail; perhaps it'd be easier to skim through. It
> seems (like most of the gerrit tools) that verbosity is thought to be an
> advantage ;-)
> [...]
>       Which is much more skim-able; of course, people may reply to such
> mails ;-) and discuss on the mailing list - presumably that causes some
> level of complexity but ...
>
>
>       Yep - on the other hand, it's nice to see it iterating towards
> something much better - clearly glueing together the mail & web
> work-flows was never going to be perfectly easy :-)

Well, I dont think the 'iterating' approach from what we currently have is
helpful here. The data source currently is these project-watch emails, which
are a not really well suited for machine parsing. So if you want to have
something completely different, instead of dissecting and reassembling these
mails, we should use gerrit stream events, which are:

- instantanteous (no delay by email polling)
- JSON objects

which can be made to whatever mail we what without fragile email parsing and
guesswork. So I would object mangling the mails beyond recognition (at least I
wont waste time on that), but keep them as 'good enough' without further mangling
until somebody hacks up a gerrit stream event solution. If you have a good
template email how that should look like in the end, we can make that a
wonderful EasyHack.

Best,

Bjoern