[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [fluid-dev] Adding C++ as mandatory dependency?
From: |
Dan Eble |
Subject: |
Re: [fluid-dev] Adding C++ as mandatory dependency? |
Date: |
Tue, 21 Jan 2020 15:43:42 -0500 |
On Jan 21, 2020, at 13:50, Carlo Bramini via fluid-dev <address@hidden> wrote:
>
> From my experience, C++ is not a good idea sometimes, especially when you
> want to maximize the performance and minimize the memory usage.
Please don't take offense at this, but demonstrating that it is possible to
write a poorly performing program in some language is not the same as
demonstrating that programming in that language is a bad idea. Even granting
that a hot spot in a C++ program might be well addressed by a tailored
solution, that does not support avoiding C++ in your entire program.
I can provide a counterexample from my own experience. I worked for 8 years in
a company whose main product performed line-rate deep packet inspection and
modification in C++. High availability, high performance, and controlled
memory use were all high priorities, and they were achieved—with diligence, but
not with great difficulty. And the developers avoided several classes of bugs
that are common in C programs, by contextually appropriate use of features like
references instead of pointers, the class system with its ability to restrict
operations performed on an object, and RAII techniques (e.g. smart pointers)
instead of the old "goto fail and try not to overlook releasing anything"
approach.
"C++ is inefficient" and similar statements can also be straw men posed by
those who feel that they would be impolite or lose face by saying, "I'm just
not interested in investing my spare time in C++ programming." I don't know
anyone here remotely well enough to claim that that is the case here, but there
are such people in the world.
Regards,
—
Dan