Re: [boost] Phoenix3 Status update.

Subject: Re: [boost] Phoenix3 Status update.
From: Joel de Guzman (joel_at_[hidden])
Date: 2010-12-27 08:36:12

On 12/27/2010 9:19 PM, Thomas Heller wrote:
> Eric Niebler wrote:
>
>> On 12/26/2010 11:53 PM, Joel de Guzman wrote:
>>> On 12/24/2010 5:48 PM, Thomas Heller wrote:
>>>>> BTW, what about compile time? Have you had any success with the
>>>>> preprocessor idea? To be honest, my biggest concern is the compile
>>>>> time of V3 phoenix over V2. Compile time is already an issue with
>>>>> Spirit2. Now we'll be adding more on top of it.
>>>>
>>>> Compile time without preprocessing is better than with phoenix2.
>>>
>>> Really? Do you have some numbers? I don't see how preprocessed
>>> phoenix3 can be faster than preprocessed phoenix2.
>>
>> I took this to mean that preprocessed phx3 compiled faster that
>> (un-preprocessed) phx2. That means that if fusion, proto and phx3 all
>> used the mpl trick of shipping pre-preprocessed headers, the compile
>> time problems would go away.
>>
>> Thomas, is that what you meant?
>
> Nope, I really meant preprocessed phoenix3 is faster to compile than
> preprocessed phoenix2 (or at least as fast as phoenix3) last time i checked.
> This was before the refactoring though. But the refactoring improved compile
> times ;)
>
> I will report real numbers once I am back home on my workstation.

If that's the case, then that's a great achievement. You and Eric have
done great work on both proto and phoenix. I'd love to see the numbers.
I'd also love to see compile tests on real world code (e.g. Spirit code).

Regards,
-- 
Joel de Guzman
http://www.boostpro.com
http://spirit.sf.net