Re: bjam rationale...

From: David Abrahams (dave_at_[hidden])
Date: 2004-12-16 22:47:08

Larry Evans wrote:
> On 12/16/2004 08:44 AM, David Abrahams wrote:
> [snip]
>>>This user-defined function can be any shell script. An example
>>>is:
>>>
>>> include $(shell echo default.imk|tee -a Makefile.d)
>>>
>>>which includes default.imk into the Makefile and records
>>>it's inclusion in Makefile.d. 'echo default.imk' can be replaced
>>>by a user defined shell script.
>> 
>> 
>> Yes, but this is inherently platform-specific.
>> 
> 
> Did you mean the tee and/or echo or the make include? If
> you meant the tee and echo, 

Yes.

> they could be replace by a
> perl script doing the equivalent for the paricular operating system.

And then we pile language on top of language. One thing we wanted to
avoid with Boost.Build was the autoconf syndrome where in order to
understand the system you need to learn four different programming
languages.

And furthermore that requires people to have Perl installed, which at
the time was out-of-the-question for some Boost developers and IMO is
still a ridiculous thing to demand of a Windows programmer who wants to
use Boost.

> Of course this would mean the install of the make system would
> have to detect this and set the appropriate links to the
> perl scripts specific to the OS, but I don't see that as much
> of a problem. If you mean the make include, then I'm stumped.

It seems to me that waving off the ways in which your hypothetical
system doesn't satisfy the requirements without having done the hard
work of building a system that does satisfy them doesn't help your case.
By the way, what case are you trying to make, anyway? Are you just
trying to tell me that my answers to the OP's questions are wrong, or is
it more than that?

-- 
Dave Abrahams
Boost Consulting
http://www.boost-consulting.com