[nanomsg] Re: Community and Maturity
- From: Drew Crawford <drew@xxxxxxxxxxxxxxxxxx>
- To: nanomsg@xxxxxxxxxxxxx
- Date: Wed, 14 Jan 2015 18:18:09 -0600
Hi Lyle,
I deploy a nanomsg-based solution at internet scale to north of 10k users for
about a year. What follows is my own opinion.
In my view the strength of nanomsg is the RFCs. They are solid and interesting
and come from a deep knowledge of networking and scalable computing.
I have found in several cases the implementation to be insufficiently visionary
in comparison. If you are running this in a very controlled environment (e.g.
in your datacenter) and you are staying close to using the published protocols
as they are intended then other than the occasional assertion failure you
should be fine.
However, in my case I’m doing a lot of things nobody intended (mostly to do
with using nanomsg as a last-mile to end clients over the internet) and it
turns out that the core implementation has a lot of drawbacks in that case. I
have improved the problem somewhat with patches in core and posts to this list
but fundamentally the effort I’ve done in core is woefully insufficient given
the problem.
I would advise you to consider Garrett’s mangos implementation as I find that a
little more malleable but for various reasons outside the scope of this thread
it’s unsuitable for my application in particular.
I think nanomsg has served me well as a stopgap solution, but it’s become clear
that for my set of problems I will need to do a cleanroom implementation from
the RFCs using a different software architecture. The open question I’m
studying at this point is whether to produce a second implementation in C11 or
to consider a more modern systems language like Rust.
Drew
> On Jan 14, 2015, at 5:54 PM, Lyle Thompson <lthompson@xxxxxxxxxxx> wrote:
>
> First of all thank you to those who helped with my previous questions. It
> shows me that the community is helpful and passionate about nanomsg!
>
> At this point, I am looking to propose nanomsg to my organization for our
> next generation product. For this, answers to the following questions would
> be most helpful.
>
> 1. How large is the community?
> 2. How active is development?
> 3. Is nanomsg reliable and bug-free enough for use in a commercial
> environment?
> 4. Is nanomsg being used in any commercial environments? If so, is
> anyone free to state which one(s)?
>
> Thanks in advance!
> Lyle
Other related posts:
