On 4/10/2019 4:21 AM, Bart wrote:
> On 10/04/2019 09:59, Juha Nieminen wrote:
>> In comp.lang.c++ Rick C. Hodgin <
rick.c...@gmail.com> wrote:
>>> Is there a Usenet group explicitly devoted to literal C or C++ compiler
>>> development? Specifically in the areas of optimization techniques?
>>
>> Unfortunately usenet is pretty much dead by this point. I doubt you'll
>> find
>> much activity on usenet groups. This might be one of the very few groups
>> that have at least a modicum of activity.
>>
>> Most communities have moved to web-based forums and Discord.
>>
>
> Which ones, things like stackoverflow? Those are extremely unfriendly
> and officious. Plus they don't tolerate any sort of discussion. You must
> ask a precise question, one that hasn't been asked before, and get only
> precise, on-topic answers. And every single thing gets voted on.
>
> Sod that.
>
> Anyway, for general discussion on language design, but also compilers,
> there is comp.lang.misc. Although usually dead, it sometimes briefly
> comes into life, as it has at the minute. It just needs someone to post
> there.
Yep.
Not really much great alternatives-wise. There are also mailing lists,
..., but they are typically specific to specific projects.
In my case, I have my own compiler which currently mostly compiles C,
and my BS2 language (sorta), a custom language (BS2), ... But, these
don't generate a whole lot of conversation.
Most notable thing about my compiler is that I have done things quite a
bit different than GCC. For example, using a stack-based IR for
statically linked libraries (though, the backend codegen itself uses
three-address-code), ...
BS2 is reasonably conservative, more or less sort of like a C/Java/C#
hybrid but without a garbage collector, and intended to have a lower
overhead and fit better into real-time applications (eg: motor control).
But, hasn't gained much ground here even in my own projects (C remains
dominant for this; and BS2 still tends to have a higher runtime overhead
than C).
I also have my own CPU ISA project (64-bit RISC variant with variable
length 16/32/48-bit instructions), in which a disproportionate amount of
effort has gone (mostly used as a VM, and still working on trying to
make a usable FPGA soft core).
...
> It's also unmoderated, unlike comp.compilers, where discussions are
> likely to be terminated if the moderator doesn't like them. But it also
> is only updated every few days at the whim of the moderator.
And also annoying as a lot of the times I have tried responding to
anything there, the posts don't get approved (seemingly just silently
dropped), so isn't really worth the effort of writing them...
But, the repetitive posts about 'Seed7' are OK?...