[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: Operationally well-tried
From: |
Ulrich Elsner |
Subject: |
Re: Operationally well-tried |
Date: |
Thu, 03 May 2007 21:30:16 +0200 |
User-agent: |
Gnus/5.1007 (Gnus v5.10.7) XEmacs/21.4.19 (Constant Variable, windows-nt) |
Robert Heller <heller@deepsoft.com> writes:
>
> Except that the Linux kernel is all plain C, not C++ -- g++ is not used
> to build the kernel.
>
> Lots of other software typically distributed with Linux is written in
> C++ and have been compiled with g++, but the OP for some reason seems to
> think these don't count.
The problem is (or I have been told it might be) that many of these
are not compiled with _one_ version of g++. For me personally, the
fact that I do not have to use exactly version x.y.z of a compiler
but can use all version x.whatever is a good sign, because it implies
that the compiler core is basically stable and the minor version
changes only affect new features or small bug-fixes. With this
world-view I can think of many examples immediately.
But question and comments suggest to me that our evaluator (?)
considers a rarely changing compiler as more trustworthy.
So, Visual C++ 6 is good.
I think the reasoning behind this is: you know the
idiosyncrasies and you can work around them.
So, if I can find some examples of software that
- uses the same version of g++
- is widely deployed, so that its stability is proven by numbers
- is preferably safety relevant (our evaluator comes from that
that corner)
I would try to use these as examples why g++ version x.y.z can
be trusted and should be used. Otherwise I'll have to try some
other approach.
Ulrich