Re: Digital Mars failures/toolset configuration question

From: David Abrahams (dave_at_[hidden])
Date: 2005-06-16 07:09:13

Aleksey Gurtovoy <alexy_at_[hidden]> writes:

> David Abrahams writes:
>> Aleksey Gurtovoy <alexy_at_[hidden]> writes:
>>> Actually it's C runtime that is missing from the online distribution,
>>> so STLPort runs are affected as well. How about we switch the tests to
>>> static linking when the toolset name ends with "-static"? That would
>>> be more general and allow us to switch back to dynamic linking (or run
>>> two set of tests in parallel) when we get the CD generously offered by
>>> Walter.
>>
>> Actually this can be done from the command-line.
>>
>> ought to do the trick.
>
> The Yahoo stripped the essentials again! 

Dang. I'll try HTML-ifying and Cc you so you can't miss it:

bjam &quot;-sBUILD=&lt;dmc&gt;&lt;*&gt;&lt;runtime-link&gt;&lt;static&quot;

> But won't anything we specify
> in the command line apply to all the toolsets we are running in the
> batch?
>
>>
>> Another possibility, thought slightly kludgey, would be to put
>>
>> {
>> local v = BUILD ;
>> if ! $(BUILD)
>> {
>> v = DEFAULT_BUILD ;
>> }
>> }
>>
>> in the toolset file. Could you try out one or both of these options?
>
> Unfortunately, no joy. It still tries to compile for/link with dynamic
> runtime. (Not that I understand how the above was supposed to change
> that).

Of course, because... it got stripped too!
It was meant to be more like:

{
local v = BUILD ;
if ! $(BUILD)
{
v = DEFAULT_BUILD ;
}
if ! &lt;dmc&gt;&lt;*&gt;&lt;runtime-link&gt;static in $($(v))
{
$(v) += &lt;dmc&gt;&lt;*&gt;&lt;runtime-link&gt;static
}
}

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