[Fixed] 5.2.12 install fails on macOS Mojave (#17805)

[IPv6Freely]

Oracle is aware of what changes need to be made, it was communicated to them already.

Can somebody from Oracle detail what the change is that needs to be made? I'd be happy to take a look at the code and submit a patch if I'm able to fix it.

[socratis]

Chris, you may have to wait until Monday (CET) to get in touch with the developers. There is also the developer mailing list, and the [url=irc://irc.freenode.net/#vbox-dev]developer IRC channel[/url].

Given the fact that you're in PST, I'd say that early Monday morning (your time) would be the best time.

[IPv6Freely]

Thank you sir, I was going to join the IRC channel today. You're right that timezones may make things difficult. I'll give it a try regardless.

[IPv6Freely]

Weird. For some reason ticket #17805 was closed with "Mojave is not yet supported".

Well DUH.

Not being supported yet is kind of the reason the ticket was opened. That's the point of being in the Apple developer program, so you can report software issues in upcoming versions of the OS. Or maybe I'm just not used to the software development model Oracle uses?

[andyp73]

I don't know what Oracle's specific support model is, but it isn't uncommon for software development companies not to accept requests through the support channel for things outside of the officially supported versions and host platforms. I work with a couple of companies whose support ticketing system only allows tickets to be raised against released versions of products.

-Andy.

[socratis]

I don't know what Oracle's specific support model is

Don't forget that models change...

It was not uncommon (and still isn't) to have open tickets for things like "VirtualBox doesn't install on Linux kernel X.Y.Z", or "VirtualBox doesn't install on Windows insider build 0xDEADBEEF". I think it was a hasty reaction, if nothing else.

And don't forget, tickets can always be reopened. As "luck" would have it, #17805 just did...

[IPv6Freely]

Thank you for the re-open!

I'm basically thinking of it as a "I’m reporting that it’s broken so you’re able to look into why so it can be fixed so that it already works correctly by the time Mojave is released to the public, instead of after" situation

I can definitely see there being a lot of situations where people who get their hands on the beta without being in the Apple Developers Program coming in and just complaining that its broken - in which case, absolutely just close those tickets!

Now, if Oracle has folks in the Apple Developer program who have the beta already (I'm sure that they do) then obviously I'm sure they're busy working on it in the back-end. That said, with Virtualbox being an open source project it makes sense to have open discussions for things like this.

The only way it would make sense to close this ticket is if the Oracle developers use a completely separate ticketing system, which if doubt - again, being an open source project.

Anyway, I wonder if the kext issue is just because Virtualbox still has some 32-bit code as has been said previously. Apple may allow some 32-bit apps to run still, but perhaps none that load kernel extensions.

[ftruter]

@ftruter
I moved your post from the thread "Kernel driver not installed (rc=-1908)". You're running 10.14 beta. Same error? Yes. Same issue? No. Read the thread from the beginning...

Sorry but my search terms didn't bring me to this thread first. Yes it is the same issue and thanks for the tip. Guess I will have to install Parallels Desktop 13 for the time being until VB updates their 32-bit code?

[socratis]

To answer both to IPv6Freely and ftruter:
VirtualBox hasn't had 32-bit code for ages. That is not the problem. Don't ask me what it is...

[IPv6Freely]

That's weird then if the previous poster who mentioned disabling 32-bit applications from running with the bootflags causes Virtualbox to not run at all is accurate.

Apparently (according to Apple) Oracle already knows what the issue is. I need to get on IRC and ask.

[Morris Meyer]

For the VirtualBox folks, the Xcode 10 toolchain is out in conjunction with this new operating system version. Could you try to use that to build the VirtualBox Mac OSX host? It seems like if the kernel extensions are in the whitelist that there might be something subtle in the loader or toolchain that it is tripping on.

--morris meyer

[IPv6Freely]

Could you try to use that to build the VirtualBox Mac OSX host?

I was actually going to try that last night, but some of the dependencies won't install for me. At least, not with macports or homebrew. Both complain about 10.14 not being supported, Argh!

Next step may be to create a 10.13 VM in Fusion (I have both) and install latest Xcode beta there and try again.

I also commented on my Apple bug report asking for more information on what needs to be modified. No response yet.

[socratis]

Morris Meyer, IPv6Freely
You don't need "the latest" Xcode to compile VirtualBox. You need a relatively new Xcode only to install MacPorts. After that, Xcode 6.2 can (and will) be used to compile VirtualBox.

See Howtos and Tutorials » Building VirtualBox » [HowTo] Build VirtualBox ≥ 5.1.2 on OSX ≥ 10.9 and its "brother" [Discussion] Build VirtualBox ≥ 5.1.2 on OSX ≥ 10.9 (you can't comment in the HowTo area, hence the need for the Discussion part).

As long as MacPorts for 10.14 is not out, you can't build OSX on 10.14. But you don't have to, not sure why this was even brought up.

[IPv6Freely]

I think we were both thinking that perhaps the latest xcode might give some clue as to the nature of the issue, thats all.

Man, they really don't like this ticket being open. "We know its not compatible with Mojave"... again, we know. But until it is, doesn't it make sense to have a ticket open until it gets fixed? What is the point of an issue tracking system if you're not tracking the issues? Just because it's not an issue with the current OS it doesn't make sense to track it? Again, maybe the Virtualbox development model is different from what I'm used to. Most of the projects I've contributed to would want it tracked so it's fixed by the time the OS is released publicly. Again that's kind of the point of the developer program.

But, if they want the ticket closed then so be it. I'll just shut up and wait, knowing that I at least tried to be useful

[ChipMcK]

it is probably a 'bean counter' thing -- those usually defy reason.