This is a great interview! Thanks a lot. I played around with Smalltalk back in highschool, so it was particularly interesting.
I do have a question, though it's rather peripheral to the conversation: How will the DLR be delivered to .Net outside of Silverlight--i.e., the main redistributable of the framework? Will that be 3.5 or a separate download like the Entity Framework?
I think you're being a bit myopic. Developers should have considerably fewer headaches as a result of all the technologies she talked about. Certainly the updated control library, in particular, will be welcomed.
I thought it was a good interview myself. Additionally, I have to say she has some of the nicest whiteboard handwriting I've seen.
This product does look very interesting to me! I, too, wonder if it will support Active Directory, so that each person in the household needs to only have one account on the network, log into any machine with the same username/password, update it in only
one place, etc.
I suspect AD might be a bit much for the target customers, but I am curious as to how authentication works. If I have a family desktop on which everyone has an account and a laptop I use for my own personal use, is there a way to indicate that the "Bryan" on
both machines is the same person? And how do you account for spoofing (since login name can be changed)?
I'm also curious as to what kind of scenarios you have in mind for businesses. Obviously, this is a win for home business, but I also think that this, perhaps used in concert with some of the Live services, would meet the needs of most small businesses. And
when someone is ready to upgrade to a more full-featured branch server or what have you, will there be any mechanisms to aid the process?
I don't really think that those two guys really understand how user accounts work in the mac. Pretending that the mac has a sinple log in/log out that puts you in an administrator account with full privileges is just a big lie to the face of the camera.
Those guys do not seem to know that OS X is a Unix like system, and for this reason it uses exactely the same model for users accounts. It works as follows: OS X as Unix or Linux uses threee different levels of permissions: - The super user or root. If a user
log in as a root, he has full provileges, full power to modify anithing in the OS. He can modify OS vital files or directories without any prompt. Well full power!!! The root is BY DEFAULT disactivated in OS X or Linux, or any other Unix. Th user need to activate
the root account manually by providing the admin password. Most of the users on mac don't even know that such account exist, only Unix users know how to activate it. - The administrator account; This is the owner account. When people install a new version
of OS X or Linux or buy a new mac, the is the default account which is created by the system. Why? Those are multiusers OS, so it needs to create at least one administrator account in order that the owner can manage the system. Of course the owner can disactivate
the admin privileges if he/she wishes. However the admin account works quite differently than windows admin accounts. On Unix a admin user can yes manage the system, set the preferences, etc, but it does not have full freedom to modify the system. If an admin
user tries to modify any OS vital files or directories, he will be prompted before. The idea is that you get the power to change things as your are the owner of the password to administrate the machine but system does not give full freedom to do anything you
want before being sure that it is reeally what you want to do. If you try to install a application that put files in protected directories, an admin will also be prompted before to do so. The difference with windows is that the admin account on Unix does not
open all doors as it does in windows, any vital change can not be done without entering a password even if your log in as admin. The admin account in windows is more similar to the super user on Unix. That means that a worm or virus will not be able to modify
any protected files or directories without the approval of the user even in an admin account. If it tries to do so the system will ask the user to prompt. In windows, under admin acccount it just go through without the user noticing that something is changing
the system. It also does mean that the admin account under OS X (UNIX) is more secure than the one in windows, because again, yes you are logged in as an admin but the system will still ask you to to enter a password if you try to make something dangerous
to the system. - The non-admin account: This the default account that is created outise of the initial adnin account. Any account which is created on OS X is by default a non-admin account, ie., with the smallest privileges. So i don't get why one of those
guys says that user accounts on mac are by default admin account, no they are not, only the orginal account created after installing the OS or starting up for the first time the mac is admin for the reason that i explained. Other created accounts are by default
non-admin with smallest privileges. That means that a user in such an account can not set the system or change any shared directories betweenn users like the Application directory. He can only change what is inside his home directory. This is quite a quick
explanation of how it works but man!! this is basic Unix. I can not just understand why those two guys seem to know very few on how accounts work on Unix and particularly on mac. Again OS X use the Unix model that i exposed. Trying to make people believe that
logging in as an admin in mac is the same as windows is just showing that he really doesn know what he is talking about. Not surprising that UAC is quite badly implemented.
But you know what we can all agree on? Paragraphs.
Thanks for the interesting interview. With the release of the VPC 2007 beta, I've really been able to play with Vista and I'm really pleased with what I've seen thus far. It's a little sluggish (which I suspect is largely due to virtualizing on a fairly
limited laptop) and there certainly are some niggling inconsistencies, but overall it's great.
One thing I do want to chime in on though is the confusing "inheritance" model in the desktop and start menu. The problem I'm having is that icons can come from either the the user's own folder or from Public (on the desktop) or Default shared folders (in the
start menu). Just by looking at it, it isn't at all clear what's coming from where and it's really rather confusing. It can also make it hard to do relatively simple things. For instance, I like to remove a lot of the marketing verbage from Start Menu entries
to keep it clean (e.g., "Windows Media Play" -> "Media Player" and so forth) and I'm finding I have to do a UAC elevation each time.
I'd like to see this move to a more prototype-based model in future versions--e.g., rather than merging a user's icons with the Default icons, give the users local copies of the icons that they can modify or remove as they see fit. Default could be seen as
a sort of "broadcast" folder for admins and new applications that install for all users. Of course, there could be GP policies to restrict user control when it really isn't desirable (e.g., public computers). Anyways, just a suggestion...
Also a few questions: Since "Documents and Settings" is now "Users" could you rename "Program Files" to "Programs" or maybe "Applications"? And why, in "Public" are the folders named "Public Documents", "Public Music", etc. instead of the more obvious "Documents",
IDEA: add a socket to the MOBO for a highspeed memory device perhaps a PCIx or perhaps a more low level - closer to the north bridge / soutbridge level...
so for example I could stick a 16 Gig flash card on it and get way faster than USB speeds!
low cost systems could have the socket w/o any card, cards could be 512 Meg 1 Gig 2 gig and so on...
USB is ok but face it... it's not as fast as the system bus by a long ways...
Let's get that stuff tight - fast as can be!
Want More Speed!!!!
Towards the ends of the interview, he talks about hardware vendors doing something similar to what you're talking about, with the added bonus that persisted data can be trusted across hibernation transitions. Of course, it sounds like the amount of flash installed
is fixed, so you can't swap it out for something with a larger quantity over time, but you still get the speed benefit of running off of PCI-X.
It was metnioned in the video that this will be controlled through registry settings, so yes this is configurable. He gave an example of being able to add LOB apps to the list of cached apps.
Well, I guess what I was really getting at is how hard is it going to be few the user to correct this optimization? Will I have to comb through the registry manually? Or will there be a software explorer similar to those in Defender?
Regarding the ability for OEMs to choose which applications are cached, is there a way for users to modify or shutoff this feature? I'm sorry if it seems rather cynical, but previous examples of letting OEMs control the software experience have been less
I love this stuff! I can seem some differences between ReadyDrive and ReadyBoost - such as wanting your hibernate file in flash that can't be removed from the machine - but otherwise, they seem similar. Are there any other differences?
I'd think the main advantage of ReadyBoost is that you can leverage it with a reasonably fast USB stick and probably just about any relatively modern computer. You can't use ReadyDrive unless you have a hybrid drive.
You are forgetting that IE is a Windows-only app. Bill Hilf likes to brag about interoperability, but talk is cheap.
You're not being reasonable at all. By providing OWA Light, Microsoft is providing a base level of interoperability. Granted, it doesn't have feature parity with OWA Premium, but it's a step in the right direction. Saying "they could do it if they really wanted
to" without having any knowledge of the team structure, codebase, schedule or budget is simply ignorant. If you read the blog post, they seem open to bringing Premium to Firefox in a future release. As subsequent versions of IE become more standards compliant,
the cost of delivering a consistent experience across browsers will shrink. It's OK to be unsatisfied, but trying to draw far out conclusions like that is just weak.
(Question: Is OWA 2007 built on Atlas? I tried googling a few sets of keywords, but couldn't find a definitive answer. It's relavent to the discussion since I know Atlas provides some abstractions to hide some differences across browsers.)
I think having SQL Everywhere databases be data-only is the right decision. Binding executable code to the format, even if done in a safe manner, could undermine the compentized model due to additional coupling. For instance, it might be harder to keep
the storage engine and query analyzer separate. The ability to pick and choose, as Anil mentioned, is definitely one of the strengths of this product.
A better solution, perhaps, would build on Richard's reasoning: make T-SQL a .Net language. This might streamline migration considerably. Although SQL Server 2005 can handle SPs in C# and VB, so migrating may actually not be that difficult. (I certainly appreciate
that some SPs are more easily written in T-SQL... though LINQ may blur this distinction.)