[OT] Modules dropped out of C++17

Joakim via Digitalmars-d digitalmars-d at puremagic.com
Mon Jun 8 20:52:24 PDT 2015


On Monday, 8 June 2015 at 19:47:07 UTC, Paulo Pinto wrote:
> That is why Microsoft is going forward with WinRT.

It is nice that they're updating the APIs, but I heard that a lot 
of it is just wrappers around old ones.  The bigger issue is that 
Windows 7 and 8 (haven't tried the preview builds for 10) both 
feel noticeably slower to me on the _same_ hardware than FreeBSD 
or linux.  I bet backwards compatibility has a part to play in 
that.

On Monday, 8 June 2015 at 19:55:34 UTC, Jonathan M Davis wrote:
> On Monday, 8 June 2015 at 19:17:03 UTC, Joakim wrote:
>> I wonder when they will realize that a clean break is 
>> necessary.  36 years is far too long for a language to keep 
>> building on top of the past.  Intel has been hurt by this with 
>> x86 recently, probably Microsoft with Windows too.
>
> There's no point in C++ having a clean break. If you're doing 
> that, you might as well just create a new language like D. If 
> C++ had a clean break, it wouldn't be C++ anymore, and many of 
> the folks who continue to use C++ are the ones who want it to 
> be backwards compatible. Arguably, if anything, languages like 
> D and Rust _are_ the clean break.

They could do something like D when it jumped to 2.0, except with 
better migration paths.  With some careful planning and automated 
source conversion tools nowadays, it doesn't have to be that bad. 
  The idea is you keep the C++ knowledge that is accumulated by 
all those devs over the years, but make breaking changes that 
allow the language to get much better and cleaner.  One of the 
main changes I'd pursue is getting rid of the preprocessor, 
though I don't know how feasible that is.

You may be right that the C++ crowd is now mostly legacy-oriented 
and wouldn't want such a change, while those who want something 
different are already moving on to D and Rust, but that doesn't 
speak too well for the future viability of C++. ;)


More information about the Digitalmars-d mailing list