3 ms·
I'm not super familiar with Apple's policies but is this really such a grave sin on OS X? The purpose of the signature, as I understand it, is mostly to assure
by pvg 7y ago
I'm not super familiar with Apple's policies but is this really such a grave sin on OS X? The purpose of the signature, as I understand it, is mostly to assure the user of the provenance of the code and, in a pinch, let Apple disable it. It's not intended as some bulletproof runtime security mechanism and it's easy to think of lots of apps that would be signed but could legitimately execute some form of unsigned code.
- oefrha 7y agoIt’s not per se, otherwise Terminal.app and iTerm2.app would be among the most sinister signed apps ever. Signatures only protect the app bundle itself, not user-supplied code or code fetched to locations outside the bundle. However, it’s bafflingly weird to include such a thing just to skip a button press or two in the installer.
- pvg 7y agoI can see how skipping the installer flow might skeeve people out but this particular bit about some signed doodad in the installer being able to launch scripts seems like something between a nothingburger and mildly curious. Just wondering if I'm missing something here.
- oefrha 7y agoYeah I don’t think this additional swipe adds much to https://news.ycombinator.com/item?id=22736608 https://news.ycombinator.com/item?id=22736608.
- foobarian 7y agoI wonder if it's just a case of an opinionated junior developer. I've been there. I've written a Windows NT console app once upon a time and there was something about the console that bugged me to no end, so I ended up rewriting it as a graphical app using my own fonts and terminal emulation. Probably wasted days doing this for some minor artifact of the UI, but it felt like a huge accomplishment at the time. Similarly here, guy or gal get it in their head that the extra click just won't do, and then find a hack around it.
- thu2111 7y agoHowever, it’s bafflingly weird to include such a thing just to skip a button press or two in the installer. Not at all. Generally every mouse click required to get an app running will slash your userbase in size by some staggering amount, like 20-50%. I can't quite recall the exact number or where I've seen this, but I've definitely heard this fact from multiple sources, including at Google. Try counting how many clicks are required to get Chrome on your system and you'll be surprised how optimised it is. Companies measure this, they're very sensitive to it of course. They want as many conversions as possible, but they can see that the more complex the install process gets, the fewer users make it through the other end. It's entirely normal for Zoom to want to simplify as much as possible.
- aequitas 7y agoIt is, from an Apple developer guidelines point of view. They have detailed docs on what situations to avoid[0] in order to elevate privaleges as safe as possible (ie: to not leave gaping holes). Granted, the best solutions they propose are not trivial to implement. Like installing a signed minimal attack surface helper that communicates over secure RPC to execute privileged functions (also don't forget to sanitise your inputs). Or having your App audited and signed so your App can access privileged Frameworks and API's (which means you have to shell out at least $100/y for a developer license for your open-source project). All in all Apple does their best to provide the mechanisms but they are not necessary easy or trivial (w're dealing with the topic of security so what's to expect). Also with every step Apple takes to bind things down there is a public outcry that they are restricting peoples freedom in installing custom applications and pushing everyone into the Appstore ecosystem. This of course doesn't excuse bad developer practices, but often it's the choice of doing things right or not being able to do them at all (never meeting your deadlines). I had the pleasure of implementing one of the suggested privilege escallation systems in a pet project[1] and it was a fun puzzle for me. But I can tell you it will be a PITA if you solve this problem under pressure as (even though Apple provides some tooling to verify everything [2]) it's really hard to figure out if all moving components are setup right for everything to work and to debug any issues. [0] https://developer.apple.com/library/archive/documentation/Security/Conceptual/SecureCodingGuide/Articles/AccessControl.html https://developer.apple.com/library/archive/documentation/Se... [1] https://github.com/aequitas/macos-menubar-wireguard/ https://github.com/aequitas/macos-menubar-wireguard/ [2] https://developer.apple.com/library/archive/samplecode/SMJobBless/Listings/SMJobBlessUtil_py.html#//apple_ref/doc/uid/DTS40010071-SMJobBlessUtil_py-DontLinkElementID_8 https://developer.apple.com/library/archive/samplecode/SMJob... https://github.com/brenwell/SMJobBless-Demo https://github.com/brenwell/SMJobBless-Demo
- pvg 7y agoThe tweet is not about privilege escalation, though, just the fact that signed code can invoke unsigned code.