Message-ID: <5408AB3F.4010803@ahsoftware.de>
Date:	Thu, 04 Sep 2014 20:11:11 +0200
From:	Alexander Holler <holler@...oftware.de>
To:	Austin S Hemmelgarn <ahferroin7@...il.com>,
	Peter Zijlstra <peterz@...radead.org>
CC:	Borislav Petkov <bp@...en8.de>,
	Måns Rullgård <mans@...sr.com>,
	Steven Rostedt <rostedt@...dmis.org>,
	Christopher Barry <christopher.r.barry@...il.com>,
	linux-kernel@...r.kernel.org
Subject: Re: OT: Open letter to the Linux World

Am 04.09.2014 19:58, schrieb Austin S Hemmelgarn:
> On 2014-09-04 13:29, Alexander Holler wrote:
>> Am 04.09.2014 16:36, schrieb Austin S Hemmelgarn:
>>> On 2014-09-04 06:16, Alexander Holler wrote:
>>>>
>>>> It's a myth that C++ ends up in bigger code than C. At least in my
>>>> experience. Especially when the latest additions to C++ are in effect
>>>> (like the move-semantics in C++11 I like quiet a lot and which you get
>>>> almost for free (by changing nothing) when you use the STL). Thread
>>>> support is now also standardized (in C++11), quiet nice to use.
>>
>>> Assuming you are writing in a standalone environment (no standard
>>> libraries), then yes, your code will usually be about the same size
>>> (unless you go way overboard with the object-oriented stuff); but the
>>> runtime is larger in almost all non-standalone environments, and there
>>> are some cases that code does end up larger in C++.  A lot of 'Clean C'
>>> (stuff written so that it compiles correctly as C, C++ and Objective C)
>>> that I have seen seems to end up larger (by about 4-6%) when built as
>>> C++ (although it usually does much worse as Objective C).
>>
>> There are always corner cases and I never would use some "Clean C" code
>> to compare sizes of C and C++. There is a whole lot of stuff you just
>> can't, shouldn't or wouldn't do when using C instead of C++.
>>
>> And just throwing in some numbers without any explanation about features
>> (like exceptions), optimizations and so on you've enabled for the tests
>> you used to get those numbers, doesn't work. ;)
>>
>> I can't really comment on what you mean with "standalone environment" or
>> "non-standalone environment", as I don't know what you mean with that.
>> But if several programms share e.g. the stuff which is in libstdc++.
>> you'll get a lot of size back when compared with C-only programms where
>> everyone invents the wheel again and again.
> By standalone environment, I mean no libraries, no libc[++], no external
> dependencies, and in the case of a lot of kernel programming, no
> built-ins.  A OS kernel HAS to be written like that, and it's easier to
> do that in C than C++.  I doubt that you have ever looked at any source
> code for Windows drivers, but Windows is written in C++, and they still
> are just as mind-numbingly insane as some of the poorly maintained,
> vendor originated Linux drivers.

I've seen drivers for Windows and for OS2/2 and DOS and FreeBSD and ...

But throwing the ball back, did you know that all Arduino SW is in C++? ;)

> Not all C is like the Linux kernel, and in fact, if you use Linux,
> probably more than half of your user-space programs were written in C.
> They use dynamic linking just like C++ programs (but often with less
> complex symbol mangling).

This thread isn't about the kernel, but some userspace program which 
does quiet a lot and which wants to do even more. I've just used one 
example I did in the kernel space to explain what I miss when I'm using 
or having to use C instead of C++.

I do understand why the Linux kernel is (still) in C and don't want to 
start a discussion about that.

Regards,

Alexander Holler
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/

Powered by blists - more mailing lists