Jean-Marc Lasgouttes wrote:
> Here is what it looks like when using a std::set for features.
> Simpler, isn't it?
Indeed.
> BTW, there is a semantics problem: in a layout file, when I do
> Require foo
> Require bar
> does it mean that I want foo and bar, or only bar?
the former. I also took car
Jürgen Spitzmüller <[EMAIL PROTECTED]> writes:
> Jean-Marc Lasgouttes wrote:
>> I do not think it matter much in terms of speed, so code should be simple.
>
> OK, I remove that.
>
>> Because map require two types, doesn't it? This is why I wrote that
>> there has to be a simpler type...
>
> OK, bu
Jean-Marc Lasgouttes wrote:
> I do not think it matter much in terms of speed, so code should be simple.
OK, I remove that.
> Because map require two types, doesn't it? This is why I wrote that
> there has to be a simpler type...
OK, but the vector does not strike my naive mind overly complicate
Jürgen Spitzmüller <[EMAIL PROTECTED]> writes:
> Jean-Marc Lasgouttes wrote:
>> Why is it necessary to handle the requires_.empty() case?
>
> It's not really necessary. I just thought I avoid the loop in this (common)
> case.
I do not think it matter much in terms of speed, so code should be si
Jean-Marc Lasgouttes wrote:
> Why is it necessary to handle the requires_.empty() case?
It's not really necessary. I just thought I avoid the loop in this (common)
case.
> What about
> using some other datastructure like map (there is
> probably something simpler for a set, though)?
Why the b
[EMAIL PROTECTED] writes:
> + if (requires_.empty()) {
> + requires_ = req;
> + break;
> + }
> + for (vector::const_iterator it = req.begin();
> + it != req.
