llvm dev - Jan 2013 - [LLVMdev] Using C++'11 language features in LLVM itself

On Tue, Jan 8, 2013 at 8:23 PM, Joshua Cranmer <pidgeot18 at gmail.com>
wrote:
> On 1/8/2013 8:30 PM, Chris Lattner wrote:
>>
>> Wow, requiring GCC 4.7 would be really aggressive, it was just released
in
>> March 2012. Call me conservative, but I was thinking that a reasonable
GCC
>> baseline would be GCC 4.4 or something (which is ~3.5 years old).
>
>
> On considering which versions get you various C++11 goodies, I'd
personally
> recommend gcc 4.5 as the baseline: that's when you can start using
lambdas
> and atomics (and a few other unicode goodies, but MSVC has yet to support
> those anyways, so the point is fully moot). Anything older than gcc 4.4
> loses you most of the C++11 features. MSVC 2010 as the baseline precludes
> you primarily from the following features:
> * raw, user, unicode string literals
> * enum class [that's 2012]
> * alignment support
> * extra constructor features
> * explicit conversion operators
> * range-based for
> * intializer lists
> * constexpr
variadic templates

David Blaikie <dblaikie at gmail.com> writes:

[snip]
>> MSVC 2010 as the baseline precludes
>> you primarily from the following features:
[snip]
> variadic templates
Irrelevant. The LLVM project discourages any non-trivial use of
templates.

On Wed, Jan 9, 2013 at 3:01 AM, Óscar Fuentes <ofv at wanadoo.es> wrote:
> David Blaikie <dblaikie at gmail.com> writes:
>
> [snip]
>
> >> MSVC 2010 as the baseline precludes
> >> you primarily from the following features:
>
> [snip]
>
> > variadic templates
>
> Irrelevant. The LLVM project discourages any non-trivial use of
> templates.

I'm not sure why you say this.

We have a fake form of variadic templates because we wanted them so badly.

We have many non-trivial templates where they make sense and provide
value... I mean, we certainly avoid them when not appropriate or when
ethere are other, simpler solutions... :: shrug ::

On the other hand, there are trivial uses of variadic templates too, so I
think we could still make the argument that they're handy to have.
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
<http://lists.llvm.org/pipermail/llvm-dev/attachments/20130109/7d1df0da/attachment.html>