Re: [Boost-cmake] Some issues getting started on Win32

From: David Abrahams (dave_at_[hidden])
Date: 2008-06-27 11:25:25

Beman Dawes wrote:
> 
> 
> On Fri, Jun 27, 2008 at 10:08 AM, troy d. straszheim
> <troy_at_[hidden] <mailto:troy_at_[hidden]>> wrote:
> 
>     David Abrahams wrote:
> 
>         Beman Dawes wrote:
> 
> 
>             I thought cmake was supposed to be robust. This is very
>             discouraging; it
>             implies cmake developers are only targeting folks who are
>             cmake experts.
> 
>             Is it possible to disable this misfeature?
> 
> 
>         I think Troy is saying that we get to control what gets cached,
>         and as
>         our system gets more refined, we'll figure out what to cache and
>         what
>         not to.
> 
> 
>     Yeah.  Don't panic.
> 
> 
> OK, I'll try not to:-)
>  
> 
> 
>         IIUC this is really the same model as autoconf and that has the same
>         issue, but it's not a big problem for people... it's just that it's
>         well-known how to blow away an autoconf cache.
> 
> 
>     Yeah.
> 
> 
> One of the lessons we have learned with regression testers is that there
> may be no people present. The scripts have to recover from errors
> without intervention of people. And adding that level of robustness
> really helps with users, too. Things "just work", and that saves endless
> trouble for users and those who support users.

We could probably introduce a "cmake; then if that fails, blow away the
cache and try one more time" sequence into the testing scripts.

-- 
Dave Abrahams
BoostPro Computing
http://www.boostpro.com