Re: [Boost-testing] depends_on question...

Subject: Re: [Boost-testing] depends_on question...
From: Gennadiy Rozental (rogeeff_at_[hidden])
Date: 2010-06-21 00:57:46

roger.wilson_at_[hidden] wrote:
> A hopefully simple question with a hopefully simple answer...
>  
> If test_suite suiteB depends_on test_suite suiteB   AND  I run with the 
> command line flag --run_test=suiteB should I expect suiteB to be run or 

Not in the current version of Boost.Test

> should I expect it to be skipped (which is what happens, presumably 
> because I've instructed only suiteB to be run so suiteA cannot run and 
> suiteB requires it to run first).

Yes. This is current logic inside.

> This behaviour seems a little odd to me.  Surely by specifying suiteB 
> specifically what I really mean is run suiteB and any dependencies 
> needed (but only those needed).  Perhaps an extra flag is required 
> allowing unspecified dependencies to run or not.

Well, this also require me to test against loops. In general Looks like 
a good idea. Can you enter a ticket for this?

Also if anyone is interested in implementing this feature, you can give 
it a shot. Otherwise I'll look into this a bit later.

Gennadiy