cespare

A contribution to an open source project is a gift with a hundred strings attached.

In most cases it's a request for more of the maintainer's time to review, fix, and merge a change which they themselves likely don't need (otherwise they would have added it). In my experience, at least 80% of the time I end up spending more time and effort on accepting someone's change than I would have writing it myself. It only becomes net-positive when a contributor has a long-lived interest in the project and turns out to be a capable maintainer with similar sensibilities; even then there is significant ramp-up time. 99% of the time this doesn't happen; most contributions are drive-bys.

P.S. It so happens that that we previously interacted on that very PR thread: https://github.com/BurntSushi/toml/pull/65#issuecomment-1919...

I hope this is considered a compiler bug.

This sense of entitlement is specifically addressed in the post.

The conclusion seems unsatisfactory. Why is lstat so much slower on their system than on the laptop?

It's frustrating that they only allow a single key. I use 4 different hardware U2F devices. Google, GitHub, Dropbox, and every other service on which I've set up U2F have always allowed me to attach all of my security keys. It seems like AWS is about 3 years behind on this stuff.

And that unique long pin definitely isn't stored in plaintext in the next column over in their database, right?

28 bits is easy. 160 bits is not easy.

I read it as an intentional play on words.

If I may engage in a little self-promotion, my tool reflex

github.com/cespare/reflex

is similar, but more general-purpose. (It's written in Go, but it's not for Go.)

GOEXPERIMENT=framepointer will be the default in 1.7, due out in a few days (see https://golang.org/issue/15840), so using tools like perf (and presumably dtrace) will be somewhat improved out of the box.

Edit: and the author mentions that at the end of the post. That's what I get for commenting before I've read the whole thing.

I don't count that against this study -- it isn't about determining the effects of a drug or something like that. If we can show that there exists one person that can detect a single photon in a statistically significant way (vs random guessing), that still seems like a fascinating result.

We push on the order of ~a billion messages / day into one of our Kafka clusters without having too much scaling difficulty.

Ah, there it is. I had read the blog post but skipped the letter.

Isn't it unusual to make an offer as an "open letter" like this? I thought these discussions generally happen behind tightly closed doors.

Related to http://passwordshaming.tumblr.com/.

Go has raw string literals.

> I do think that SL made the right decision though.

I assume you mean LambdaConf, not StrongLoop.

> Never again will you have to hunt through your tabs to pause your music,

> Supports media keys (play, pause, stop, next, previous)

A relatively recent Chrome feature that many aren't aware of is the ability of extensions and apps to use Chrome- or desktop-wide (global) hotkeys. So if you're using the Google Play Music extension, you can, for example, hook it up to your media keys and it will work even when Chrome isn't focused. Go to chrome://extensions/ and scroll down the Keyboard shortcuts to learn more.

(I'm sure that there are many other reasons people prefer desktop apps, but this was the big one for me.)

Yep, gofmt defines and applies a standard formatting that uses tabs for indentation. Using tabs means that everyone can use their own preferred tab stop in their editor while operating on the same code.

It's confusingly worded, but the OP turned down an AWS offer; they don't work at Amazon.

I saw the same thing a couple of months ago with a key/value datastore that synchronously writes each update to disk. Switching from ext4 to xfs fixed periodic 100ms+ latency spikes.

It's nice to read an article with a bunch of charts that all have a y-axis minimum of zero.

BTW the name for this is "false balance".

https://en.wikipedia.org/wiki/False_balance

You can do better -- fmt provides a way to give a dynamic width or precision as an int argument.

  fmt.Fprintf(os.Stderr, "%*s %s\r", paddingWidth, "", legend)

I'd further wonder whether a separate padding calculation is needed at all (too lazy to go read the source on Github at the moment).

Here's a collection of offenders:

http://password-shaming.tumblr.com

It's a kernel bug that was reported to Apple years ago.

Fortunately, it seems that it might finally be fixed in El Capitan:

https://github.com/golang/go/issues/6047#issuecomment-135591...

So a "trade canary", then.

But the NaCl key is randomly generated for each connection.

Interesting. A couple of questions that come to mind:

- Any rough benchmarks vs. TLS? Or even just back-of-the-envelope math/reasoning behind the claim in the opening paragraph: "without the overhead of TLS".

- Instead of generating 24 PRNG bytes for each message to use as the NaCl nonce, why not use the sequence number each message is assigned anyway?

This was also one of my very first projects when I started writing Go a few years ago.

https://github.com/cespare/go-tetris