A betterC modular standard library?

[rikki cattermole via Digitalmars-d]

On 20/12/2016 10:32 PM, Ilya Yaroshenko wrote:
> On Tuesday, 20 December 2016 at 09:12:55 UTC, Timothee Cour wrote:
>> @ilya: regarding -betterC 's stated goal to bypass druntime:
>>
>> besides static linking (which can be inefficient), why not use runtime
>> shared libraries to distribute mir (eg GLAS) to C clients ?
>>
>> IIRC, from a single C++ program, I am able to dlopen & dlsym & run D
>> functions defined in shared libraries built with different compilers
>> (eg dmd and ldc) even if they use GC (and therefore dmd's and ldc's
>> druntime) ; pseudocode:
>>
>> // main.cpp
>> ```
>> void main(){
>>   dlopen, dlsym, and run "void fun1()" from libfun_dmd.so
>>   dlopen, dlsym, and run "void fun2()" from libfun_ldc.so
>> }
>> ```
>>
>> That would provide the needed isolation without suffering from
>> disadvantages of static libraries (eg bloat); compared to the betterC
>> alternative, you'd have access to all of druntime/phobos and not have
>> to reinvent the wheel
>
> Because I do not have control on phobos, i can not distribute my copy of
> phobos with required patches like normal cache size estimation
> (core.cpuid is broken). Why not to fix core.cpuid?  Its source code is
> not manageable comparing with mir-cpuid: it is faster to write a module
> from scratch for this case. DUB package appears when I want it, LDC
> release may appear after 6-10 months (depends on distribution).

Of course you can bundle your own Phobos.
If you put a std.* module in your path your build will be working 
against that instead. If you don't mess around with sc.ini it will 
override the default for that compilation.