In these dark times, it's inspiring to see that a country as insignificant as Australia can demonstrate to us how things can always get worse.

By passing a law mandating that encryption must be broken, in secret (like the US's National Security Letters), at the demand of the Government, the two biggest parties have colluded to destroy Australia's tech sector.

This is the same government that has been whining endlessly about using Huawei LTE equipment in Australian infrastructure "because it might be secretly compromised". Now the same is true of Australian equipment, by law.

My favourite part of all this is how there will be firmware updates for devices sold in Australia, in order to comply with the new law. How well do you think those backdoors will be secured? How thoroughly do you expect them to be tested, given Australia's population of only 25 million?

How can any Australian company expect customers to trust them now?

2codepaths1function

Microsoft's Azure team understands.

Why the fuck do coding blogs insist on using themes with a 600px content column, then use code samples that are 3x wider than that? The whole reason I have a widescreen monitor is to not _have_ to scroll, jackass!

I wish the U.S. didn't "pasteurize" eggs, because without that, eggs are shelf-stable, which would mean I could keep them in a basket on my countertop and label it "npm"

Today, I learned the shortest command which will determine if a ping from your machine can reach the Internet:

ping 1.1

This parses as 1.0.0.1, which thanks to Cloudflare, is now the IP address of an Internet-facing machine which responds to ICMP pings.

Oh, you can also use this trick to parse 10.0.0.x from `10.x` or 127.0.0.1 from `127.1`. It's just like IPv6's :: notation, except less explicit.

Oh JavaScript... can you seriously not even increment the exponent of a float without barfing?

*siiiiiiiiiiiiiiiigh*

protected static new readonly

// just no

Fuck MS, why couldn't you update the NuGet API URL when NuGet updated? The warning on nuget.org states,

"This package will only be available to download with SemVer 2.0.0 compatible NuGet clients, such as Visual Studio 2017 (version 15.3) and above or NuGet client 4.3.0 and above"

It says nothing about using the V3 endpoint, so if you''re like me and updated NuGet to 4.5 and still got nothing but

"NU1101: Unable to find package Foo.Bar. No packages exist with this id in source(s): https://www.nuget.org/api/v2/"

...then you'll be very confused until it strikes you that there might be a new API version. Even if MS doesn't want to deprecate the V2 API just yet, it would be awfully nice to just state on the frickin' site that not only do you need NuGet >= 4.3.x, but also the correct feed URL.

$_DEITY knows how many dev-hours have been lost to this shit.

> be me, working on small addition to enormous feature branch
> build system in flux due to reorganization started a month ago, not quite solid yet, but mostly works
> f_branch gets master merged into it sometime last week
> bossman makes "minor" change to build system and edits master to match
> doesn't merge changes into f_branch
> bossman goes on holiday for a week
> no permission to merge master changes into f_branch
> linter barfs
> npm barfs
> build server barfs
> mfw I can't even deploy to our testing environment

I never had this happen using TFS...

Opus is an amazing codec, but did Soundcloud really have to switch to it with a bitrate of 64kbps? even 80kbps would have been worlds better.

Bandwidth isn't _that_ expensive, even post-neutrality.

the people in Ops
all have space heaters, but we
don't have the power

Seriously though, building management needs to turn up the heat by like 3°C. And install new breakers. And fix the shitty wiring.

Just created a CLUSTERED INDEX -- knowingly and intentionally -- for the first time. I feel like a frickin' sorcerer.

Visual Studio's Test Explorer is a piece of shit.

Maybe the user wants to repeat the last test run without rebuilding? No can do. Maybe the user didn't add or remove any tests, but just needs to rebuild without running test discovery again? Nah. Maybe the user just needs to discover tests from THE ONLY ASSEMBLY WHICH WAS REBUILT? Too frickin' bad.

A 120-second turnaround (30 to build, 90 to discover) just to _start_ a test run is bloody atrocious. Especially when VS decides to run test discovery twice in a row for no given reason.

*sigh*...

I'd use ReSharper's runner, but unfortunately it isn't capable of running xUnit v2 tests when you've designated a custom XUnitTestFramework.

GodDAMN, C# 7.0 is so ridiculously feature packed. I can pattern-match inside a predicate on an exception filter. Want to catch ONLY NHibernate's exceptions caused by a SQL timeout? Boom:

catch (GenericADOException adoEx) when (adoEx.InnerException is SqlException sqlEx && sqlEx.Number == -2) { return Failure("timeout"); }

You know what's more fun than debugging a SQL stored prodecure?

Debugging a SP which CATCHes all errors and instead returns an error code. Because exceptions are scary...

Office manager just mandated that our standing desks would have a "cable pouch" installed at the rear for cable management. My cable tray **was** the neatest of everyone in dev, now it's unscrewed on the floor and all my cables plus the laptop charger are hanging loose, because notmyjob.jpg and contractors generally DGAF.

Oh, and they didn't install it flush with the rear of the desk, presumably because they didn't want to take off my laptop stand. So it's right in the way of my feet when I'm sitting.

Nothing that I can't fix with a screwdriver in my own time, fortunately.

Fuck off OneDriveSetup.exe, nobody asked you to install anything. This "i7" is only dual-core, and I need both of them to run my code, kthx.

It looks like Microsoft are back to their old tricks, specifically the DirectX 9 naming scheme. Naming releases after Northern Hemisphere seasons and repeating words never gets old and/or confusing!

Who's looking forward the "Winter (2018) Creator's Update"?

Using ReSharper is like becoming enlightened, or de-brainwashing oneself to see true reality. Of my entire dev team, I'm the only one who can see the fnords!

Unused identifiers, badly sorted modifiers, unused property setters, redundant `this`/namespace, redundant casts... Surely if they could see them too, such code would not survive!

Lodash, Rimraf, Grunt, Gulp... I'm still not convinced that our frontend guy isn't just playing Pokémon Go all day

Goddamn, Windows' idea of symlinks is completely broken. It's like they faked it at the UI level, but if your build process wants to copy the file? Too bad, it's not real so you can't copy it.

Javascript is a lot easier when you remember that there's a convenient REPL that is never further than an F12 away