Re: [Boost-testing] Dated results

From: Aleksey Gurtovoy (agurtovoy_at_[hidden])
Date: 2005-12-16 04:04:29

Stefan Slapeta writes:
> Aleksey Gurtovoy wrote:
>  >
>  > Some runners can only afford to run regressions once a week or
>  > so. Also, even one-month-old results for an exotic platform could
>  > be of an interest if they are the only ones, although they
>  > probably should be archived to another page or something along
>  > these lines.
>  >
>  > Further thoughts are welcome.
>
> The reason for my post was that I'm currently not able to get a
> feeling of what is right and what is not in boost when I look at a
> test matrix that has become very large.

Understood. We are hoping to address that with richer view options; in
any case, usability feedback like this is very much appreciated.

> I don't know what could help, but maybe it's just introducing
> another layer on top of the current matrix where each toolset is
> mentioned only once and which refers then to the detailed matrix.

This definitely can be one of the possible views.

> Another option could be ordering the matrix by the actual test run
> date from left to right and skipping the platform ordering....?

This one would be harder to implement, but how about filtering out
everything that is more than, let's say, a day old? It would be a much
lower-hanging fruit and is in fact already on the plan.

Thanks for the feedback!

-- 
Aleksey Gurtovoy
MetaCommunications Engineering