Re: [Boost-testing] Re: TARBALL SNAPSHOT IS BROKEN! Was: Locale support with Codewarriow/Darwin (again)

From: Rene Rivera (grafik.list_at_[hidden])
Date: 2005-03-10 21:43:40

Aleksey Gurtovoy wrote:
> Rene Rivera writes:

> Does what? Uploads logs to server? It's a one-liner.

And for the next feature.. another one-liner, and another.

>>So there's no point in continuing to "extend" regression.py if we
>>can just switch to an already proven equivalent.
> 
>   ^^^^^^^^^^^^^^^
> 
> To switch, you have to transfer all the logic of 'regression.py' to
> BuiltBot server side.

Some of it. Parts of it have better equivalents in BuildBot. And other 
parts are the same logic I, and others, use in our test scripts (not 
regression.py).

> Then everybody needs to switch to the new test
> running infrastructure. 

Yes. Incrementally.

> The cost of implementing the whole thing is
> significantly higher, although it's definitely more exciting to work
> on than fixing the mundane problems of the existing tools.

I _am_ fixing an existing tool, regression.py, collect_and_upload.py, 
run_tests.sh, run_tests.bat. And yes, I'm also fixing other tools, like 
BBv1, just not as visibly.

And it has nothing to do with excitement. This is something I have a 
personal need to do, i.e. I need for my business. I can't use 
regression.py for my needs. So I chose to do some extra work, above *my* 
requirements so that it could be used by Boost. In the mean time I have 
a 3 hour interruption, to track down a problem on your machine.

Honestly Aleksey, you ask me to explain how BuildBot can be better. And 
when I do, you belittle my answers, and the work done by the creators of 
BuildBot. I'm not sure how to take it, but it's getting a bit 
frustrating :-\

-- 
-- Grafik - Don't Assume Anything
-- Redshift Software, Inc. - http://redshift-software.com
-- rrivera/acm.org - grafik/redshift-software.com - 102708583/icq