Re: Bjam core tests

From: David Abrahams (gclbb-jamboost_at_[hidden])
Date: 2003-05-14 06:22:11

Vladimir Prus <ghost_at_[hidden]> writes:

> We were discussing this issue with Dave and I think we've reached a 
> conclusion. But I'd like to make sure we all agree to it. The proposal is for 
> a simple rule
>
> All non-trivial changes to Bjam core must come together with tests.
>
> Does everyone agree?

Of course. I did supply tests.

> The motivation is that jam/bjam itself is not very good tested, and if we 
> continue adding untested features, bjam will only degrade. The more direct 
> issue is that recently, typechecking facilities were added to bjam, and I did 
> not seeing the exact description of semantics, or any tests. As result, I had 
> to look up in code to understand how it works. A simple tests (which I 
> already comitted) would be sufficient to understand everything. 
>
> (There's some test in class.jam, actually, but it's not executed by
> default 

"not executed by default?"

You must've missed this:

local rule expect_derived2 ( [derived2] x ) { }
^^^^^^^^^^

expect_derived2 $(d) ;
expect_derived2 $(e) ;

> and is too high-level to understand how new feature works).

What's high-level about this?

Actually, your test is not such a good demonstration because type
checking code should generally not ECHO, but should return a message
as my example did.

-- 
Dave Abrahams
Boost Consulting
www.boost-consulting.com