[zeromq-dev] [PATCH] Improved response to socket violations

[Pieter Hintjens]

On Fri, May 20, 2011 at 6:17 PM, Martin Lucina <mato at kotelna.sk> wrote:

> There's always a tradeoff. I recall the discussion from last year about
> whether or not ZeroMQ is a library for "Real men" or "Little old ladies".
>
> Personally, I think you can have "Fastest messaging ever" or "safest
> messaging ever", but not both. Of course, feel free to submit a patch which
> accurately detects such conditions while keeping the lock-free and high
> performance nature of the system.

This really isn't about performance (there's nothing in my patch that
affects that, afaics) but about not wasting the time of core devs on
handling problems that aren't really there.

Again, if you have a better answer, please do provide it. The problem,
I'll restate: people report assertions, Martin and I spend time
investigating, helping people make test cases, and after some effort,
it turns out to be socket violations. It happens often enough to be
annoying, and I'd like to solve that, not argue over philosophy.

There is already improved documentation. People do _not_ read the FAQ
(from web site stats, we know this).

So, constructive ideas?

You cannot seriously be suggesting that "real software" should be
painful to use and painful to support.

-Pieter

Ps. and BTW, there are other assertion failures, such as when
socketpairs run out of space, which could also be done more helpfully.