Live data from Hacker News

The Day AppGet Died (2020)

keivan.io

121–130 of 132 posts

Re: The Day AppGet Died (2020)

#121
I remember reading this when it was first posted and sensing that there was some disconnect between the author and Microsoft.

The reframing of the hiring as not part of a wider career at Microsoft seems exactly as it sounds - "ok, you don't have to do the stuff we've suggested ... But we still want AppGet".

And then this:

> We talked about some of our options, but in general, I thought everything went well.

It's quite clear that it didn't go that well, isn't it?

I'm not defending Microsoft here or suggesting that this was the exact outcome the author was seeking, but I sense a bit of naivety in interpreting what Microsoft were driving towards.

Ultimately they clearly decided the effort of a grounds up redesign / rewrite was more valuable than bring him along for the ride.

Re: The Day AppGet Died (2020)

#122

I dunno, getting inspired and copying all the ideas, deciding that they can do a good enough job internally instead of hiring Keivan doesn’t sound like a bad thing in and of itself. It’s the stringing along that makes my blood boil. After reading things like this, I’m thinking, what I’d do differently? Maybe have them reimburse all my travel upfront. No money, no meetings. Also, bill all the meetings and conversation…

>I dunno, getting inspired and copying all the ideas, deciding that they can do a good enough job internally instead of hiring Keivan doesn’t sound like a bad thing in and of itself. It just completely kills the motivation to make any kind of system tools like this. If your tool fails, you have wasted your time. If it succeeds, the OS vendor will kill it overnight with a free knock-off, so you have also wasted your t…

You do these tools for yourself and to scratch your own itch. As long as it fulfills its function, you get value. The fact that someone else may want to use it or the big vendor would like to adopt it should be secondary at best.

If you yourself don’t need such a tool and you also don’t want to do it as an exercise of the art, then don’t do it. Spend your time on something that’s gonna bring value to you instead.

Re: The Day AppGet Died (2020)

#123
post #103

Earlier quoted context omitted.

>I dunno, getting inspired and copying all the ideas, deciding that they can do a good enough job internally instead of hiring Keivan doesn’t sound like a bad thing in and of itself. It just completely kills the motivation to make any kind of system tools like this. If your tool fails, you have wasted your time. If it succeeds, the OS vendor will kill it overnight with a free knock-off, so you have also wasted your t…

You could release your code with a less permissive license (say excluding companies with a 1 trillion dollar market cap) so they at least have to play ball with you.

If these companies think they can get away with unlicensed use, they will. It’s you going broke because of a lawsuit, not them even if they lose.

Re: The Day AppGet Died (2020)

#124
post #106

Earlier quoted context omitted.

The GPLv3 might be enough to do that, for some. It was sufficient concern for Apple, who dumped Samba, bash, and gcc. Microsoft's CBL/Mariner distribution of Linux would be the first place to look for GPLv3, and and impacts upon their patent portfolio (I haven't read up on the patent provisions). "Apple, a user of GCC and a heavy user of both DRM and patents, switched the compiler in its Xcode IDE from GCC to Clang,…

LLVM is also more modular. The front-end and back-ends are well decoupled and it’s possible to work on them separately. Useful for a company that has to target two CPU architectures.

I have heard that many optimizations that llvm can make are impossible in GCC.

I barely grasp yacc, but I understand that a more cohesive design, less burdened by history, can be greatly effective.

There is a group that is trying to rework Linux in such a way that it can be compiled by llvm clang. I wish them the best.

Re: The Day AppGet Died (2020)

#125
post #49

Earlier quoted context omitted.

> You want separate logins for every service? In short: yes. In long: I want separate logins for different domains. In no universe should Minecraft share a login service with Azure cloud engineering. And I shouldn't have to log in to use offline applications locally. And teams should support different logins for many accounts, like Slack figured out ages ago. By forcing changes to the login pattern for negligible to…

You can create as many Microsoft logins as you like. If you want to separate Minecraft and Azure, go ahead. But there's no reason this requires separate authentication technology or infrastructure.

>But there's no reason this requires separate authentication technology or infrastructure.

Only if you are Microsoft replacing working login system with a new one costs zero.

But as a user you have no reason to change and many reasons not to do it, especially if you experienced already issues from the Skype account migration issue and you seen the news where giants ban your account for shitty reasons , so now I might lose my MS account access if my son somehow gets banned in Minecraft.

But you will say is more beautiful from an engineering POV to have one account to rule them all, my response is that is not as easy, I might already have 2 Skype accounts, 2 Minececraft accounts and 1 MS account that I use(MS might have created some other ones behind my back), it is ugly not beautifull.

What would have been beautiful maybe would have been from those smart devs to keep the old email and password login work and move their login server to Azure ,Windows and .Net if they are allergic to whatever Mojang uses, that is beautifull when dev changes help the users or don't screw with themn, changes that help devs are just for dev ego.

Re: The Day AppGet Died (2020)

#126

Earlier quoted context omitted.

You can create as many Microsoft logins as you like. If you want to separate Minecraft and Azure, go ahead. But there's no reason this requires separate authentication technology or infrastructure.

>But there's no reason this requires separate authentication technology or infrastructure. Only if you are Microsoft replacing working login system with a new one costs zero. But as a user you have no reason to change and many reasons not to do it, especially if you experienced already issues from the Skype account migration issue and you seen the news where giants ban your account for shitty reasons , so now I might…

I don't really care, but I can only imagine that your opinion of what is "beautiful" doesn't carry much weight with the folks that have to build and maintain this shit.

Re: The Day AppGet Died (2020)

#127

This is why you use, at the very least, a version of the GPL. But he used Apache. So he basically did Microsofts job for them for free.

Like someone else mentioned, this is not a source code issue. I think 90% of the people who got pissed back then thought Microsoft stole the code: get this, they didn't steal anything. One is C++ and the other is C# so it is literally impossible for a license to matter here. This would have been solved only with software patents. But I can very easily guess you're against software patents, aren't you? Or only when th…

That's pretty fervent. I think you're worked up about what you wish the parent had said. You should have left it at the first paragraph.

Re: The Day AppGet Died (2020)

#128
post #43

Please don't use any of these. Use scoop - it just works https://github.com/ScoopInstaller/scoop No admin needed for almost all packages, "scoop update *" to update your system, authoring packages is very easy.

For anyone new to scoop, it is seriously The way that package management should work, in Windows. Apps install as the user (so no admin needed), and scoop uses a “current” symlink to point to the current version of the apps you have installed. Switching between versions or even hard-linking to a specific version is trivial. I was a little disappointed to see that some folks are ditching it for winget. While we definitely don’t need to go back to the Wild West “install things wherever you want” days, Windows needs a good package management ecosystem that works responsibly.

Re: The Day AppGet Died (2020)

#129
post #33

Earlier quoted context omitted.

> requiring you to have a Microsoft account for future versions of Windows 11. How quickly people forgot about mandatory iTunes activation... > They're still the company that force-upgraded users to Windows 10 This is response for all those cries what MS should take the security of the OS seriously. No wonder what after a decade of disabled Windows Update on the endpoints (to conserve resources, to not to receive WPA…

> mandatory iTunes activation Did macOS ever have mandatory activation? Because I have used it for a long time and have never needed that. You need some way to get a new OS version, which was buying a new macOS version on a disk (back in the day), then on the app store, and then it was a free upgrade from the app store. But you never needed an account to run the actual OS.

iOS, not OSX

Re: The Day AppGet Died (2020)

#130
post #23

Earlier quoted context omitted.

> For people like me who had WU disabled ... Yes. This is the reason it is so hard to disable WU now. YOU are one of those reasons. > forced approach only worked on the people who had auto-update enabled Also I want to hear what kind of magic should they used to do that on people with WU disabled.

I would've been one of those people, if I'd ever had a security breach. At this point I've been saved so much time and hassle not putting up with Microsoft's nonsense I could spend a month dealing with a stolen identity and I'd still come out on top. Of course I'm not one of the numpties that runs untrusted JS in my browser or installs rando programs.

> Of course I'm not one of the numpties that runs untrusted JS in my browser or installs rando programs.

You, as the other people in this thread, are forgetting/don't think what 90% of the Windows userbase struggle to distinguish between a click and a double click.

Personally I'm not fond on how and where MS directs Windows for the last 8 years, but it's becames totally understandable when you take the sheer number (and ignorance/stupidity) of the userbase.

Post reply on HN