Re: [boost] [rfc] I/O Library Design

From: Jeremy Maitin-Shepard (jbms_at_[hidden])
Date: 2007-06-18 17:37:18

"Michael Marcin" <mmarcin_at_[hidden]> writes:

> Jeremy Maitin-Shepard wrote:
>> 
>>> It would be more complex and slower to compile, but it as some
>>> advantage.
>> 
>> I think that any sort of complex expression template framework would
>> be totally out of the question, because the increase in compile-time
>> would make the library so inconvenient to use that it would just not
>> be worth using at all.
>> 

> That's a pretty bold statement.

Well, in fairness, I did not exclude the possibility of a "simple"
expression template framework being suitable.

I certainly stand by the argument that long compile times and large
generated code size are not at all acceptable for a facility like text
formatting that is intended to be used very widely and in even the
simplest of programs.  You can put a Boost Spirit parser in a separate
file that you try not to rebuild very often, but you cannot do the same
with basic use of text formatting.

If it turns out that some sort of expression template framework doesn't
have these problems, then all the better, but in my experience with such
frameworks, those issues are hard to avoid.

-- 
Jeremy Maitin-Shepard