The Official 10.7 request thread
Ok Snow Leopard hasn't even shipped yet but we know what it's going to have so it's time to look beyond 10.6 and peer into the palantir. This thread is about technology that needs further fleshing out or new technology that should be added to 10.7.
I presume 10.7 should be a paid upgrade so there are going to need to be significant enhancements to warrant spending the money. Snow Leopard looks to be a better performing and more streamlined OS. It's going to become progressively harder for Apple to move people off of their installed OS.
My initial entry:
1. New UI with a healthy dose of Resolution Independence.
a. I expect laptops w/hirez screens and mobile devices to leverage RI most but so will
larger LCD with higher PPI.
2. Global metadata enhancements that allow for tagging.
a. I'd love to see Apple spearhead something like OpenCL for dealing with customizable
metadata that can be read on multiple platforms and filesystems.
3. ZFS or BTFS support for modern fs
a. With snapshots, checksums, pooled storage and de-dupe
4. Security virtualization
a. Sandbox everything in its own vm. Any rogue process can be quarantined at any
time.
5. Core UI
a. Finally Apple allows safe hack free theming.
Feel free to add your own.
I presume 10.7 should be a paid upgrade so there are going to need to be significant enhancements to warrant spending the money. Snow Leopard looks to be a better performing and more streamlined OS. It's going to become progressively harder for Apple to move people off of their installed OS.
My initial entry:
1. New UI with a healthy dose of Resolution Independence.
a. I expect laptops w/hirez screens and mobile devices to leverage RI most but so will
larger LCD with higher PPI.
2. Global metadata enhancements that allow for tagging.
a. I'd love to see Apple spearhead something like OpenCL for dealing with customizable
metadata that can be read on multiple platforms and filesystems.
3. ZFS or BTFS support for modern fs
a. With snapshots, checksums, pooled storage and de-dupe
4. Security virtualization
a. Sandbox everything in its own vm. Any rogue process can be quarantined at any
time.
5. Core UI
a. Finally Apple allows safe hack free theming.
Feel free to add your own.
Comments
Oh no...
Mr Ireland <Agent Smith voice>
No need to elaborate...we already know.
Tablet.
Mr Ireland <Agent Smith voice>
No need to elaborate...we already know.
Tablet.
10.6 didn't hit the shelves and we're making out list for 10.7 - that makes me say "oh no...". Can't we be happy with what we've got? Is that too much to ask??
10.6 didn't hit the shelves and we're making out list for 10.7 - that makes me say "oh no...". Can't we be happy with what we've got? Is that too much to ask??
Oh yes!
I'm thrilled with Snow Leopard but Apple is going to be talking about 10.7 a year from now. Which means they are already in the process of designing the next featureset.
Snow Leopard isn't hard to wrap your mind around. It's an extension on what Apple has been working on in Leopard for some time.
Oh yes.
Voice Dictation -
a. Not only is voice dictation a must for people with disability but it represents the next
step above Voice Controls and should work well as a desktop app and iPhone app.
1. New UI with a healthy dose of Resolution Independence.
a. I expect laptops w/hirez screens and mobile devices to leverage RI most but so will
larger LCD with higher PPI.
Yes as it means that hi res laptops should be able to use non-native resolutions for people who can't read small text easily.
2. Global metadata enhancements that allow for tagging.
a. I'd love to see Apple spearhead something like OpenCL for dealing with customizable
metadata that can be read on multiple platforms and filesystems.
I'm not sure how OpenCL would help with metadata but some kind of cross-platform standard for meta-tagging would be good. This unfortunately won't depend on Apple but Microsoft.
3. ZFS or BTFS support for modern fs
a. With snapshots, checksums, pooled storage and de-dupe
If it has enough advantages sure. I think faster, more reliable SSD storage will alleviate a lot of the problems we have and it may be enough to simply evolve HFS+.
4. Security virtualization
a. Sandbox everything in its own vm. Any rogue process can be quarantined at any
time.
Nope, I don't think this would work well. You'd constantly get requests for authorization every time an app decided to do something outside the sandbox, which could get very annoying quickly like in Vista. Having the option would be good but not by default. The iphone sandbox model causes more problems than it solves such as leaving users without copy and paste for 2 years.
5. Core UI
a. Finally Apple allows safe hack free theming.
Yes, I'd like to see that. OS X's interface is getting very tired as is the iphone's and not tired because it's bad but because it's old.
All I really want to see in OS X will hopefully arrive in 10.6. Finally processing on the GPU, 64-bit, fixes for all those niggling problems in 10.5 such as the disk cannot be ejected when it's in use. If 10.6 is good enough, 10.7 doesn't need to come for a good 2 years afterwards.
Nope, I don't think this would work well. You'd constantly get requests for authorization every time an app decided to do something outside the sandbox, which could get very annoying quickly like in Vista. Having the option would be good but not by default. The iphone sandbox model causes more problems than it solves such as leaving users without copy and paste for 2 years.
Isn't that what Ivan Krstic wants to do? Bitfrost
http://radian.org/~krstic/talks/2007/auscert/slides.pdf
Slide 55
It's really only for untrusted code so I'm thinking that with code signing (if I've got my security lingo down) become defacto something akin to Bitfrost takes over for everything else (meaning external data sources)
Isn't that what Ivan Krstic wants to do? Bitfrost
http://radian.org/~krstic/talks/2007/auscert/slides.pdf
Slide 55
It's really only for untrusted code so I'm thinking that with code signing (if I've got my security lingo down) become defacto something akin to Bitfrost takes over for everything else (meaning external data sources)
It could certainly help prevent some issues but the user is always given the ability to override security measures and so in the best case, it comes down to social engineering.
I liked the suggestion that on install, the user would be asked what the app should be allowed to do - in the case of OS X, this would be on the first execution attempt - but a user may not know what an app can do for example Delicious Library using your isight to read barcodes in which case the user must be subjected to another prompt or the app maybe crashes because the developer didn't test the app under every security restriction.
Once an app convinces a user to give it filesystem access, it can do damage, maliciously or accidentally. Even a basic script embedded inside an app (or more likely a saved file of that app) with a interpreter runtime can wipe your filesystem of all your important data in a matter of seconds - not overwrite them that quickly but recovering even deleted items on OS X is difficult and time-consuming.
I think OS X's security model works well enough. Keep the critical parts of the OS locked down, give the user freedom to do what they want without restrictions or prompts and encourage routine backups.
An improvement I'd like to see is in the installer to see what items will be installed and where. The 3rd party app Pacifist lets you do this. Beyond that, I don't think there is an adequate solution that will prevent all possible attacks.
It would be good to have warnings that notified users though such as a warning that an app was modifying the system folder or perhaps a library folder that isn't commonly modified as well as a warning about making a change to particular folders. Perhaps if more than 100 files are modified in the home folder within a certain time.
It's still not perfect because in the latter case, an attempt to batch modify over 100 images would be blocked by a warning. But then it could say something like never warn me when batch encoding images with this app.
I do know that today's computer security is draconian and consumer unfriendly and if Apple can deliver the type of impact (which is ironic because his ideology seems to be about not impacting users with security concerns.
Instead of protection from executing
untrusted code, Bitfrost protects the
machine while executing untrusted code.
Wow. Instead of trying to force developers to write fail safe code we'll give them a sandbox to crap all over and force the user to restart and crap again and again and again.
If you noted the Keynote you saw a hint at code signing that will drive Cocoa apps to be legitimately signed apps before they are allowed to even install on OS X.
This means they would have some sort of mechanism for Packaging that requires testing on the application before it's packaged and available for download.
If Apple is going to invest all their time and efforts into LLVM and the Clang project they will definitely work at covering those edge cases for C/C++/ObjC/ObjC++.
You know with ZFS being dropped from the latest builds i'm going to go out on a limb and state that 10.7 will be ushering in a new filesystem.
As cool as ZFS is it seemed like Apple quickly embraced it and then lost interest. Sun was on the block a month ago and Apple could have bought the whole kit and kaboodle but they didn't and my thoughts are this.
1. Apple has no desire to play in the Enterprise space.
2. ZFS was too Enterprise focused
3. Patent issues with Netapp and the Oracle buyout don't help
4. Apple may be looking at a homegrown filesystem that scales from mobile devices to mid level biz needs.
Dominic Giampaolo came to Apple seven years ago to work on the Filesystem team and work with Spotlight team. He was instrumental in creating the BE filesystem which leveraged metadata. I think this is where Apple wants to go. A lighter filesystem with modern features like checksum and perhaps snapshots and CoW and more.
If Apple has be cultivating and baking a filesystem for say the last 4 years it would appear that 10.7 would be the ideal target. They've added 64-bit to OS X from kernel up and eliminated legacy PPC support.
Bringing back Themes from OS 9 would be great and is something I've always wished for since switching to OS X. This would alleviate people wanting a different colored interface throughout the OS.
2) Linux emulation to allow popular Linux GUI applications like GIMP, Pidgin, etc to run from binaries without any recompile. This would allow Sun's version of Java for Linux to run seamlessly.
3) Multiple boot environments like OpenSolaris. If an update nuked you installation, you can roll back to the previous one and restore any subsequently added files with Time Machine.
4) Ability to open a Finder window from Spotlight the same as any other application so I don't have to take my hands off the keyboard. Why it doesn't now allow this, I'm not sure.
5) Touch screen interface (if this isn't here already).
6) Application-specific volume controls, like Vista.
7) Fully configurable keyboard shortcuts like in GNOME.
8) APIs to allow applications to be displayed full screen. (ex Firefox has a full-screen mode on Windows, Linux and OpenSolaris, but not on OS X).
9) FTFF. Now that the Finder is (apparently) 100% Cocoa, it's time for Apple to add features to it (ex, copy and paste full directory/file path) to make it more competitive with Windows Explorer and to give it more sensible keyboard shortcuts (ex: ENTER to open a file instead of renaming it).
more sensible keyboard shortcuts (ex: ENTER to open a file instead of renaming it).
Yuck. There's command-o for that.
It would be nice if Apple relaxed control slightly. If you really want return to open files instead of renaming, why not simply have that as a Finder preference? That way you can have enter=open and I can have enter=renaming. Another case in point - safari 4 tabs: the beta had them in the title bar (interesting idea but quite horrible in terms of UI guidelines etc.). Some people hated them and some people love them. But, now in the final release they've been removed completely - again, why not give people the option?
I would also love to see folders that display what's inside of them Or some quick view option that shows you what's in the folders.
In Leopard every folder looks exactly the same. it's fine if you have just a couple of folders but if you have an extensive collection it's tedious looking for that specific file because there's no way to tell you what's in the folder except for the label.
I want Resolution Independence as well! Resolution independence would be my number one request for OS 10.7.
I would also love to see folder's that display what's inside of them Or some quick view option that show's you what's in the folder's.
In Leopard every folder look's exactly the same. it's fine if you have just a couple of folders but if you have an extensive collection it's tedious looking for that specific file because there's no way to tell you what's in the folder except for the label.
Dude! Just because a word ends in "s" doesn't mean it needs an apostrophe!
folder's -> folders
show's -> shows
look's -> looks
And yes, all those mistakes did make your post hard to read.
Dude! Just because a word ends in "s" doesn't mean it needs an apostrophe!
folder's -> folders
show's -> shows
look's -> looks
And yes, all those mistakes did make your post hard to read.
relax dude! it was just a spelling mistake, I don't see how it made my post difficult to read. I edited it out if it makes you feel better.
1) Apt-get-like installer or (OS X app store if you want to call it that) to allow users to easily search for, view the properties of, install, update, or delete software, including open-source software.
2) Linux emulation to allow popular Linux GUI applications like GIMP, Pidgin, etc to run from binaries without any recompile. This would allow Sun's version of Java for Linux to run seamlessly.
3) Multiple boot environments like OpenSolaris. If an update nuked you installation, you can roll back to the previous one and restore any subsequently added files with Time Machine.
4) Ability to open a Finder window from Spotlight the same as any other application so I don't have to take my hands off the keyboard. Why it doesn't now allow this, I'm not sure.
5) Touch screen interface (if this isn't here already).
6) Application-specific volume controls, like Vista.
7) Fully configurable keyboard shortcuts like in GNOME.
8) APIs to allow applications to be displayed full screen. (ex Firefox has a full-screen mode on Windows, Linux and OpenSolaris, but not on OS X).
9) FTFF. Now that the Finder is (apparently) 100% Cocoa, it's time for Apple to add features to it (ex, copy and paste full directory/file path) to make it more competitive with Windows Explorer and to give it more sensible keyboard shortcuts (ex: ENTER to open a file instead of renaming it).
Sorry, but I have built custom kernels on Debian for the past 9 years and managing Experimental and Sid systems all the way through.
Apt-Get Installing is not a godsend.
I've filed roughly 40 broken package upgrades in the past month on Debian. There is no way in hell you want an apt-get upgrade system.
There is a new Installer coming but this notion that Debian or Redhat's RPMs have this clean install/uninstall structure is ludicrous. The policy issues are changing constantly within both distributions. Compound that with Ubuntu, CentOS and other branches of Debian doing slightly different policy rules than Debian only produces more crap when they merge back the quite often in-flux of Ubuntu -x updates to countless packages.
Never mind the fact that OS X is about to begin it's actual inaugural Openstep 5 release [12 years after the merger] with it's well defined and known system for file structure management that is completely different than a traditional System V UNIX based OS, Snow Leopard will provide a much cleaner installation structure for Cocoa packages.
It doesn't mean Apple is going to care about local packages via Fink/apt/dpkg which pollutes your /usr/local structures to make it all one happy family.
Regarding shortcuts for Filesystem paths you can write a Service in Automator to build yourself a system of ~shortcut alias to map to where ever you want to go. Of course you want Apple to do that for you, but seeing as they won't know how your network is mapped out with User Accounts [assuming you have more than one system] ~/Username =>/Network/Subnet/Group/Hosts/Users/ or if you'll have a network wide deployed /LocalApps exported mounts to keep your local machines lean and more don't expect that type of addition to happen until 10.7.
If you don't know what I'm describing then you've never worked in a large deployed network of NeXTSTEP systems. It was silly how well mapped and managed NeXT's corporate systems were designed to allow one to access any resource anywhere in the world.
What's with this Linux Emulation? GTK+ is being ported to OS X. Qt is there already. Run your apps natively in those toolkits. When GTK+ team rolls out GNOME 3.0 I'd expect most of that work in the toolkits for work on OS X to happen. It's already on Windows. Why in the hell would Apple waste cycles on making GTK+ apps emulate in OS X outside of XDarwin when the toolkit is coming to Apple via the GTK team?
I'm not even going to address Keyboard Shortcuts until you've seen 10.6 and you've addressed your issues with the app developers you want to extend that access to you.
Why you don't build Services with Keyboard Shorts via Automator.app seems something I imagine some developer will do and charge for it.
If you're asking for Apple to embrace GRUB 2 and thus give back the changes to the GPL I will just say, DREAM ON. It would be nice but other than debug kernels for Devs you won't be seeing 10.6.1, 10.6.2, 10.6.3, etc, boot options.
Sorry, but I have built custom kernels on Debian for the past 9 years and managing Experimental and Sid systems all the way through.
Apt-Get Installing is not a godsend.
You do realize that I said "apt-get like", don't you?
I've filed roughly 40 broken package upgrades in the past month on Debian. There is no way in hell you want an apt-get upgrade system.
The difference is that Linux distributes application files all over the filesystem. OS X does not. It's all in the same .app folder, which means 99% of the issues with apt-get go away.
There is a new Installer coming but this notion that Debian or Redhat's RPMs have this clean install/uninstall structure is ludicrous. The policy issues are changing constantly within both distributions. Compound that with Ubuntu, CentOS and other branches of Debian doing slightly different policy rules than Debian only produces more crap when they merge back the quite often in-flux of Ubuntu -x updates to countless packages.
Seeing as I'm not a beta tester and a lot of Snow Leopard details are under NDA, how was I supposed to know there was a new installer?
Never mind the fact that OS X is about to begin it's actual inaugural Openstep 5 release [12 years after the merger] with it's well defined and known system for file structure management that is completely different than a traditional System V UNIX based OS, Snow Leopard will provide a much cleaner installation structure for Cocoa packages.
Again, how was I supposed to know this?
It doesn't mean Apple is going to care about local packages via Fink/apt/dpkg which pollutes your /usr/local structures to make it all one happy family.
Yeah, I agree that is annoying, which is why I asked for a new system-wide installer feature.
Regarding shortcuts for Filesystem paths you can write a Service in Automator to build yourself a system of ~shortcut alias to map to where ever you want to go. Of course you want Apple to do that for you, but seeing as they won't know how your network is mapped out with User Accounts [assuming you have more than one system] ~/Username =>/Network/Subnet/Group/Hosts/Users/ or if you'll have a network wide deployed /LocalApps exported mounts to keep your local machines lean and more don't expect that type of addition to happen until 10.7.
That's a poor excuse. There's no reason Apple can't limit the keyboard shortcut functionality by reserving some keyboard shortcuts or keeping them off limits.
If you don't know what I'm describing then you've never worked in a large deployed network of NeXTSTEP systems. It was silly how well mapped and managed NeXT's corporate systems were designed to allow one to access any resource anywhere in the world.
I'm a home user with only a single account and a single Mac at home (I have an OpenSolaris system on my second computer) so this is really not the first thing that comes to mind when I think of keyboard shortcuts.
What's with this Linux Emulation? GTK+ is being ported to OS X. Qt is there already. Run your apps natively in those toolkits. When GTK+ team rolls out GNOME 3.0 I'd expect most of that work in the toolkits for work on OS X to happen. It's already on Windows. Why in the hell would Apple waste cycles on making GTK+ apps emulate in OS X outside of XDarwin when the toolkit is coming to Apple via the GTK team?
The idea is to take Linux binaries and run them natively, like FreeBSD does with apparent efficiency.
I'm not even going to address Keyboard Shortcuts until you've seen 10.6 and you've addressed your issues with the app developers you want to extend that access to you.
Like I said, I'm not a paid up beta tester, and I don't have access to NDA'd knowledge.
Why you don't build Services with Keyboard Shorts via Automator.app seems something I imagine some developer will do and charge for it.
Because I'm too lazy? I suppose it's something I could set up after doing a bit of reading, but the point of this thread is to put together a wishlist.
If you're asking for Apple to embrace GRUB 2 and thus give back the changes to the GPL I will just say, DREAM ON. It would be nice but other than debug kernels for Devs you won't be seeing 10.6.1, 10.6.2, 10.6.3, etc, boot options.
No offence, but do you know anything about OpenSolaris? OpenSolaris is under CDDL, not GPL (2 or 3). Also, the multiple boot environments are driven off ZFS snapshots. The Grub configuration merely takes those snapshots and allows you to boot into them.
With ZFS and DTrace incompatible with the GPL, do you really think Sun would contribute their snapshot feature to the GPL and allow Red Hat and Novell to sell it? Dream on!
Seeing as Apple can take such a feature from Sun's CDDL code anytime they want, it would be a killer feature that Apple could use to sell OS X to enterprise customers.
OS X 10.7 Aslan