Re: [Boost-build] Python port development

Subject: Re: [Boost-build] Python port development
From: Stefan Seefeld (seefeld_at_[hidden])
Date: 2010-07-08 14:12:39

On 07/08/2010 03:08 PM, Robert Ramey wrote:
> Stefan Seefeld wrote:
>    
>> On 07/08/2010 12:58 PM, Robert Ramey wrote:
>>      
>>>    I realize that this is off topic - but really curious as to why
>>> you've selected python rather than C++
>>> for this effort. Would you care to list you're reason(s)?
>>>
>>>        
>> May be I'm misunderstanding what you have in mind, but to me the
>> answer seems obvious: Makefiles (or Jamfiles or SConstruct files,
>> etc.) are essentially scripts (with a domain-specific language) that
>> are interpreted. How would you do that with C++ ?
>>      
> bjam is a script interpreter.  Seems to me that C++ is meant to
> be the best there is for this kind of job.
>    

What makes you say that ?

>> Or are you thinking of inventing a new such DSL and writing the parser
>> for it in C++ ?
>>      
> That's what bjam does now.
>    

True, and it is arguably not particularly good at it.

>> I believe one of the biggest advantages of moving to
>> Python is that it is a well known language,
>>      
> aaa... and C++ is not?
>    

I'm talking about the scripting language, not the programming language 
with which the parser is written. For a build tool I couldn't care less 
what the parser is written in, as long as the scripting language is easy 
to use.

>> with well-define semantics.
>>      
> and C++ semantics are not well-defined?
>    

Again, that's beside my point.

>> Writing a DSL on top of that is much less work (and will
>> be easier to extend, as well as apprehend by users), than any custom
>> language (including the language currently used in bjam /
>> boost.build).
>>      
> So the "new" bjam would have a different language than the current v2?
> That's new to me.
>    

We are not talking about bjam here, we are talking about boost.build. 
It's all the build system (scripts) that would move to a different 
language, IIUC.

As far as I know, Rene is rewriting bjam (in C++), but for entirely 
different reasons. I just hope both (Rene and Vladimir) will collaborate 
or at least coordinate on this.

     Stefan

-- 
       ...ich hab' noch einen Koffer in Berlin...