> Wow that is really damning, trashing user security in trade to remove a single click

As someone who has had to develop and maintain a similar web-to-desktop bridge I can tell you that this one issue was responsible for around 90% of my company’s total support requests, despite only being a small feature in a optional addon in one of our main products.

For businesses just trying to keep their customers happy, this particular one click is a very real problem.

I can absolutely symphetize with people trying to come up with workarounds.

Did your company try to write an FAQ page on how to accept the double-confirm dialogs in all major browsers (with screenshots) to maybe reduce your "90% of support tickets"? Does your ticketing software redirect you to (or display) an FAQ page that matches the ticket title?

I've come to understand how features like these get built, but I've also come to understand that people that use software are a lot more resilient and savvy than we think.

If 90% of your support tickets are about getting through a standard double-confirm, patio11 would probably recommend increasing your pricing to limit your paying customers to a pool that probably won't have much more trouble with that.

Ahhh, sweet naivety.

The people who end up calling support often aren't the ones paying the bill, for starters. That's definitely the case for Bluejeans and Zoom.

It also doesn't matter if you make a great FAQ page. Majority of dissatisfied people will never see it. Majorly because they won't call support but instead complain and grumble locally, the second biggest portion because once sent towards FAQ by support.... They won't follow it.

Only the tiny sliver of most dedicated will follow up long enough to reach the FAQ.

That is interesting. What were the support requests, how to remove the confirm step, or what to do if you denied it but didn’t mean to? Or something else?

Most desktop-browsers have in the name of security made it exceptionally hard to accidentally launch external programs through this mechanism.

We’re talking software engineering phd can’t complete it without hand-holding hard (true story!)

So normal users definitely don’t understand nor manage to navigate the dialogs presented by the browser to produce a “successful” outcome.

In the past we used this mechanism to “automatically” provide configuration-data a desktop component, so that it could call back to our application. And our users just didn’t manage to configure it.

In the name of security, browsers made one path so hard to use, without considering what people would then develop instead.

And here we are now. Oops!

I have to call BS on this. Are you claiming that users can't complete a single prompt of "[Your browser] needs to open an external application to follow this link. (Decline) (Launch Application)"? That seems really unlikely. I've done enough user testing to, at least anecdotally, say with some certainty that this is not true.

It’s a two-dialog process (allow website to use external protocol & what external program should be used for this protocol), with intentionally confusing wording making it easy to choose the wrong choice (disallow) if you don’t read thoroughly.

Unless you already know what to do it’s fairly unintuitive.

Most users don’t even know the difference between a single click and a double click.

Expecting them to even know what an external protocol is, or why it should be launched at all is completely unreasonable.

And this is why the web browser vendors need to simply disallow this behavior. Websites seem to think they seem to engage in awful behavior to compete with each other. If the browsers just block it outright, then everyone will be on a level playing field.

Thats kinda what happened here, Apple made the user confirm that they were going to take an action on an app on their computer and Zoom built functionality in on the app to bypass it.

No, you’re misunderstanding me. Websites should not be able to talk to localhost over HTTP. Browser vendors should eliminate the back channel.

There are no “developer terms” that developers have to abide by for the Mac.

If they're notarized there's around a hundred pages of terms and conditions you have to agree to. Although I'm not sure this gets in the way of any of them except on one of the blanket ones that Apple keeps intentionally vague.

Yes if 6 == 100.

A sibling reply posted the link.

https://developer.apple.com/terms/apple-developer-agreement/...

Can you expand? I'm not sure what your comment means.

You said there are “about 100 pages of terms and conditions you have to agree to”. There are six pages.

I literally just went through them at work in preparation for Catalina. They're hidden behind the developer account stuff you need to get your certs registered.

Ahem.

https://developer.apple.com/terms/apple-developer-agreement/...

And you are not forced to abide by the developer terms to release a program for the Mac just as I said. Despite all of the HN conspiracies, you can develop for and release code on the Mac without Apple’s permission. You don’t have to abide by those terms.

The Zoom and BlueJeans apps are both signed, which indicates they're part of Apple's Developer Program, and thus bound by its terms.

If you want to distribute an unsigned app and guide users into bypassing Gatekeeper for it, by all means, do so... but that's not what's happening in this case, nor is it particularly common due to the intentional hoops they make you (or more accurately, every single prospective user of your app) jump through.

And that still doesn’t contradict my statement that you can distribute apps for the Mac without being in the developer store....

Wait a while. Holding one's breath not recommended.

For those who don't want to (or cannot, due to the nature of their application) use the Mac App Store to distribute software, the requirements will only continue to get more specific until (to the extent possible) all executable code and resources are notarized and signed with an identity.

<Insert Perry the Cynic rant about unsigned code - "What the hell is wrong with you!?">

This same prediction has been going on since 10.6 - over 7 versions ago. How do you propose that Apple forces code signing on programs that run on top of a VM like the CLR or JVM? How do you propose they enforce it for programs run using a scripting language? The best they could do is force signing on the runtimes.

But my point still stands. Today on July 9th 2019 you are not forced to be part of the developer program to distribute apps on the Mac. Despite all of the pollyanish the sky is falling type that has been going on for over a decade.

Not to put too fine a point on it, but those examples don't pass muster.

- If you're not using .NET, the CLR doesn't affect you, and although Microsoft has done well with ,NET, I wouldn't necessarily expect Apple to make Redmond's job easier.

- Java is much the same boat, and is perhaps in even worse shape as it used to be included by default in macOS releases but now isn't.

Read: security nightmare.

- From 10.16 on, scripting languages also aren't included by default. This seems less adversarial than the situation with Java, but for things like Homebrew, it's a stumbling block they will need to overcome.

https://discourse.brew.sh/t/mac-os-deprecating-system-script...

Apple introduced the Mac App Store over a decade ago. Since then, conspiracy theorists have been predicting that Apple will force all apps to be signed.

Are you predicting that Apple will disallow all scripting language runtimes and all VM based development environments? So if these same predictions have been wrong for over a decade - and still aren’t happening with 10.13, exactly when will this happen?

As far as Apple not including (outdated) versions of various scripting languages or Java - neither does Microsoft. That hasn’t been a major impediment to adoption.

Sigh, you're just not getting it, sorry to say.

I have NO TROUBLE imagining that Apple will continue to tighten the screws on this, enforcing signing through Developer TOS and requiring MAS apps to pay for distribution certs.

Direct download isn't going away, not after all the work that's gone into securing it, but if you think you can sell an app off your own site without giving Apple some identifiable info about who you are and what your code does, prepare to be disappointed.

Runtimes won't be disallowed, just that you (the user) are responsible for installing them and keeping things updated.

Oh, and for record, my reference to "Perry the Cynic" is no accident...he literally invented how code signing works.

https://weblog.rogueamoeba.com/2008/03/07/code-signing-and-y...

https://red-sweater.com/blog/514/development-phase-code-sign...

http://patft.uspto.gov/netacgi/nph-Parser?Sect2=PTO1&Sect2=H...

http://patft.uspto.gov/netacgi/nph-Parser?Sect1=PTO2&Sect2=H...

So you realize you’re kind of arguing against your point? He made a prediction that still hasn’t come true over a decade later.

And citing the patent office isn’t helping either. Every company patents everything they can.

Direct download isn't going away, not after all the work that's gone into securing it, but if you think you can sell an app off your own site without giving Apple some identifiable info about who you are and what your code does, prepare to be disappointed.

Well today you can. As you have been able to do since the info-Mac archives since before the World Wide Web existed. So unless you can bring back some proof from either your time machine or visiting some other world in the multiverse, I would rather talks about facts as they exist today.

And code signing still won’t stop you from being able to run code that runs on top of a VM or scripting languages without them being signed and you won’t have to do the ctrl-click bypass.

Why is it wrong for Apple not to bundle extra runtimes (scripting/JVM) software that increases the attack surface? Should they also start back bundling Flash?

> Well today you can. As you have been able to do since the info-Mac archives since before the World Wide Web existed. So unless you can bring back some proof from either your time machine or visiting some other world in the multiverse, I would rather talks about facts as they exist today.

Watch WWDC 2019 Session 701, you'll learn something.

https://developer.apple.com/videos/play/wwdc2019/701/

> And code signing still won’t stop you from being able to run code that runs on top of a VM or scripting languages without them being signed and you won’t have to do the ctrl-click bypass.

It is easy to do this? No, in many cases I'd expect it to be a serious P.I.T.A, but it's unquestionably the right move going forward.

https://mjtsai.com/blog/2019/06/17/notarizing-command-line-t...

That has nothing to do with distributing the programs that run on top of VMs/runtimes. The operating system only sees the JVM/CLR as an executable. Even if that has to be signed, there is no way of enforcing the programs that run on top of them to be signed.

Today, on July 9, 2019, yes you are not forced to be part of the developer program.

Apple has announced that is changing very soon[0] and you attacking everyone who already knows this as 'conspiracy theorists' is kind of insulting.

0: https://developer.apple.com/documentation/security/notarizin... - "Beginning in macOS 10.15, notarization is required by default for all software".

You can only have software notarized as a member of the developer program.

My statement was very clear “developers are not forced to be part of the developer program.” Meaning that unlike ios, there are ways to distribute your app without signing it. “Notarization required by default” != “there is no method to distribute unsigned apps”.

https://www.google.com/amp/s/eclecticlight.co/2019/06/07/not...

Catalina still runs apps which haven’t been notarized or even signed, including those built after 1 June 2019. But you may find them more complex to run, and they don’t of course benefit from any of new security protection unless they’re signed and hardened.

You left out the portion of the article that says only apps you built yourself don't need to be signed/notarized.

Apps distributed over the Internet, like, you know, the ones we're talking about, must be notarized according to your own source.

What part of Catalina still runs apps which haven’t been notarized or even signed, including those built after 1 June 2019. But you may find them more complex to run

Is difficult to understand? In Catalina just like in the current OS, there is a built in method for the end user to bypass code signing for any app. The user can choose to run unsigned third party code.

The article states that code you create doesn’t have to be signed and you don’t have to go through the “complex” process to run it.

Third party code forces you to go through the “complex” task of ctrl clucking.

I'm fairly certain you have to agree to some to join the Apple Developer Program, which is (sort-of) required if you want people to be able to run your app.

You don’t have to be in the developer program ftp distribute your app on the Mac.

Technically? No.

It'll largely refuse to run ("App can't be opened because it is from an unidentified developer") if it's not signed via Gatekeeper, though.

There's a procedure to bypass that, but it's hardly user-friendly. https://support.apple.com/kb/ph25088?locale=en_US

Yes control click is really complicated.

First, the dialog doesn't give any indication it can be bypassed in that fashion. Second, users should rightfully be suspicious of "just bypass the security!" install instructions - especially non-technical ones.

Because users got a lot of security installing Zoom and BlueJeans since it was signed....

This isn’t unique to these two. Dropbox does some ungodly things when installed on the Mac....

How is it trashing user security?

By running a web server with the ability to circumvent installation? It could almost be considered a backdoor

User willingly installs their software, they are not backdooring it. There are plenty of services running on every computer, including TCP servers and web servers. If you're going to call them backdoors, you've got a long list.

If I think I have installed your software and it's secretly running a web server with the ability to reinstall itself, I am calling it a backdoor. I think that list is pretty short, but perhaps I am wrong.

Look at what just happened with Zoom. The web server could be used by a malicious third-party to gain access to the system.

I don't think I understand the security implications of this either. Seems distasteful, but how could it be exploited?