CppCon 2017: Charles Bay “The Quiet Little Gem in Modern C++: ＜system_error＞”

—
—
Amongst the loud fanfare of C++11 arrived this quiet little gem of ＜system_error＞, with std::error_code and std::error_condition born from the heart of boost::asio. With Committee input they evolved for inclusion into the C++11 Standard, providing consistent and extensible ways for applications to expose platform-specific error-codes, platform-independent error-conditions, and rely upon an open-ended design that permits future extensibility for message reporting; and even internationalization of logs and user-facing events and conditions.

More than half a decade later, we most unhappily find that the motivation and intended use model for std::error_code and std::error_condition are still not well understood; even in circles eagerly embracing features and idioms encouraged by the latest C++ Standard revisions. This may be somewhat expected, as all things “error” tend to permeate system-wide design-decisions and influence the processing metaphor; how algorithms compose conditional success-or-failure branching; and create consistency challenges across APIs (both internally, and for interoperation with third-party libraries).

We discuss the features and design of ＜system_error＞ and its intended use; make recommendations regarding API design; and propose possible extension mechanisms for module-specific concerns, instance-specific messages (with embedded value-reporting), and internationalization.
—
Charles Bay: F5 Networks Inc.

Software developer with 25+ years experience in large-scale and distributed systems in performance-sensitive environments including real-time processing, performance visualization, embedded systems, time-sensitive processing of large data sets, hardware status-and-control, and instrument/process/sub-assembly monitoring.
—

*-----*
*-----*

[BretKuhns]

Gets into actually using system_error at 50:32 and wraps up at 1:02:17 .

[sekaita]

the horrific microphone feedback stops after a few minutes.

[mikeleist]

I wish I had read all of the other's comments before wasting an hour of my life. Advice to presentation author : Remove 80% of the slides, SLOW DOWN, breathe and focus on the salient points... and we are not dumb, we do not to be told 100 times over about something. I do not consider myself a C++ expert, though I am a C/Unix expert, and I found this presentation to be totally overblown - surely 10 minutes would have sufficed to describe this stuff, or 20 minutes to describe it properly!!

[hanneshauptmann]

A good talk should have nice content and at least a simple and nice presentation on the beamer.
I think this talk has some content, but I dont get anything out of it. Its sometimes hard to follow,
the pdf on the screen is sometimes REALLY overloaded.

A good talk gets supplemented by the visuals, in this case .... well, you know already I guess.

I hope you were able to pick up some tips, Mr. Charles Bay.

[llothar]

This happens when 4chan kids grow up and become experts.

[mingchiisuen]

Quite a LITTLE gem... indeed, in this talk.

[maylermartins]

Very annoying echo sound. Hard to concentrate in the talk with all this noise.

[NbyEdge]

I have been anticipating the videos of some other talks. Is this really the one you chose to upload?

[simplyshow]

Don't watch this video for its technical content - unless you have a tonne of time on your hand. A six minutes talk has been stretched to an hour and ten minutes! Utter lack of respect for people's time. If Mr Bay wants to rant on and post that video that is fine. But the talk should make it clear. Not all of us have that kind of time on hand.