Re: [Sablevm-developer] Are we ready for 1.1.4?

[Grzegorz B. P. <ga...@de...>]

On (11/05/04 23:00), Chris Pickett wrote:
> The following sentence is redundant, otherwise it looks fine.
> 
>    So, in general, the "staging" code is very robust.

Hmm... it surely is a repetition, though we might want to keep it to
emphasize that the code IS very usable.  This is something, I think,
that really is different in SableVM development model.  And I also
think that we *want* as many people as possible to use 'staging'.

Personally, I'd keep it.

> At some point soonish (either before the end of the summer or before 
> 1.2.0) I plan on getting some regression testing stuff into staging (it 
> will require the user to download benchmarks separately).

Yeah, I am going to set up some UML (User Mode Linux) env. on our new
server so that we could:
a) securely build real, 'dist' tar archives - current staging snapshots
   are, as David noticed, simply tar.gzs of tree checked out from
   subversion,
b) run nightly any tests we might think of, I hope on more than one
   architecture; UML might be preventing us from getting reliable
   *performance* results though, but all the rest should work just
   fine.

So when you have some regression testing suite it will surely be put
into use.  I imagine we'd want to have there things like:
a) SPEC bencharks
b) BTF
c) Mauve
d) Ashes (are they usable currently?)
e) ...
and we'd have "alarm" emails sent to the ML if nightly testing shows
any anomalies in the output on any architecture.  Nice, ain't it?

I know what you think...  and no, I am not just brainstorming :-)
Though these things surely won't be operational tomorrow.

Cheers,

					GBP

-- 
Grzegorz B. Prokopski <ga...@de...>
Debian GNU/Linux      http://www.debian.org
SableVM - LGPLed JVM  http://www.sablevm.org
Why SableVM ?!?       http://devel.sablevm.org/wiki/WhySableVM