Re: [review][fixed_string] Andrzej's feedback

From: Andrzej Krzemienski (akrzemi1_at_[hidden])
Date: 2019-12-05 14:07:01

czw., 5 gru 2019 o 14:39 Vinnie Falco via Boost <boost_at_[hidden]>
napisał(a):

> Folks, we are way overthinking the behavior of resize() over
> capacity(). It throws for a very simple reason. The most obvious and
> common optimization for highly concurrent network programs is simply
> to avoid the global allocator. And the obvious and common mitigation
> for resource exhaustion attacks is to limit the amount of data that an
> I/O handler is allowed to process.
>
> Using a fixed string as a temporary buffer for an algorithm satisfies
> both of these goals, and it does so without the caller having to write
> a bunch of code to check sizes in advance.
>

Thanks you. This is the kind of information that I was after.

I would like to know more. Given that fixed_string does not have operator+,
as well as the description above, I am inclined to think that it is not
treated as a string, but as a buffer of characters. Next question is, why
is a vector with fixed capacity not enough for this purpose? Do you ever
have a need to call `.find_first_not_of()` in your use cases?

Regards,
&rzej;