So the waiting is over for now. I had a bunch of expectations for the new release of Mac OS X and now that the cat is more or less out of the bag, it‘s time to compare my wishes with what will be delivered.
The feature I have been waiting for a long time is the correct handling of metadata on the Mac. And it seems as if Apple has made good use of the time they spent on thinking about the right strategy. Take a look at Spotlight to get an idea, of what Apple wants to deliver.
The good news is that Apple is setting the knife on the biggest shortcoming of OS X so far: lack of metadata (data about data) support in the file system. While HFS+ was built for storing and accessing metadata from the beginning, OS X did not make much use of it and the underlying Darwin layer did more or less ignore metadata stored in the files. The biggest problem is backups: it is almost impossible to back up a OS X system with the onboard tools. You have to use at least on tool out there to get rid of that. But I believe backup up a system (especially syncing the data to an already existing backup copying just the changes) should be a core feature that must work without buying expensive tools that introduce other bugs or limitations. In Tiger, the UNIX Userland (the Darwin command line programs) all gain access to metadata. So you can then copy a file with cp or create an archive with tar and retain the complete structure. On Panther, files with metadata get silently destroyed by these tools and this was a headache not only for novices. The introduction of access control lists (ACLs) makes administration for systems with many users a lot easier. The good integration in the GUI receives my applause as well. I just hope it will be stable from the beginning.
Another welcome addition is the introduction of the launchd subsystem that takes care of starting, stopping and restarting background services, unifying access to programs both from UNIX and Mac OS heritage. This might allow for less booting at installation time as programs might be able to register and start their drivers and background processes in a well-defined way, removing the need for a clean bootstrap phase. Technically, there is almost no need to boot OS X at all when installing software, but it is done by many programs still (even Apple‘s). This has to go away and the launchd approach seems pretty sane to me. If launchd does to launching what lookupd did to name lookup on OS X, admins will get more sleep.
The second biggest advancement in my point of view is the introduction of the Jabber protocol to iChat. I know, most people will primarily thrilled by the audio and video conferencing features (I am too), but the move to the IEEE standard XMPP protocols means much interoperability, more security and customizability for smaller and bigger organizations.
Mac OS X Server includes an “iChat Server” which I guess is just a well-integrated jabberd process with a nice admin GUI. This server can handle your domain and allows for secure connections with SSL. The new iChat will register on your server instead of AOL‘s and will only open connections on the Internet if you want to talk to somebody outside of your networks. This makes iChat usuable in corporations that do business and don‘t want their discussions go unencrypted on the network. Also, the move to Jabber eliminates the absurd limit of 150 people on your buddy list which is already my biggest problem with the AIM system.
Interestingly, iChat has used the Jabber protocol for a long time already: the Rendezvous chat function works using Jabber instead of AOL‘s Oscar protocol. That‘s why I never feared Apple would not do the move. But it‘s good to know it‘s actually happening and that they are entering the IM Server market as well. This is another good reason getting Mac OS X which has been the best UNIX integration I have ever seen already. Tiger will improve on this tremendously.
Another area that Apple addresses perfectly is Weblogs. They are going to throw in first class RSS support in Safari with a really well thought-out user interface. Applause again, the team really hit the point. Use Google for “old”, “static” information. Use RSS for searching up-to-date, fast-changing information from the sources YOU define. This is really great.
What is even better is that Mac OS X Server packs a Weblog Server (it is actually Blosjom) as well, making OS X the ideal tool for class rooms. I have done this a while ago: weblogs for students are a perfect way of keeping everybody busy reporting on the progress. What was missing was the integration with the user database for simple authentication. Being able to sign on to login, file sharing, instant messaging, weblog services and all the other cool services in OS X is breathtaking. Setting up something like this can take weeks on a Linux box and in the end you‘ll find the integration is probably not that well. With OS X Server this can be done in minutes. Additionally, Apple packs automatic cluster building so that a second machine can take over at any time the primary server fails. Clap clap.
I have been using Konfabulator for a while and although I was not that impressed by it‘s overall speed, some tools have been quite handy like the iChat Bezel and the cool Weather widget. Now, Apple integrates Konfabulator in OS X under the name Dashboard and we‘ll see a wealth of new, fancy widgets sprawl once Tiger ships. The Exposé-style appearance of the widget is just the right way to do it: I have played around with the widgets a lot and found their co-presence usually annoying. But once you wanna use them it important to be able to access them fast. The “put it on top and remove it fast afterwards” approach is a winner in my point of view.
I could applaud a couple more of the additions, but I wanted to turn my focus more in the lesser discussed aspects and leave the praising to the Macintosh weblog world. But I also need to look at what is still missing in this new OS.
First let me say that I expect much more to come to Tiger than what was presented at WWDC. Apple focused on new technology that desperately needs fast developer adoption if it should succeed: Dashboard is useless without Widgets and Spotlight needs applications to open their internals to the search APIs. This will happen, I am sure. RSS is already here and Apple‘s support will promote Weblogs in public to what the net underground already knows they are: first class citizens in the web world. This is good for the infrastructure of the web and will enhance the usability of web sites tremendously, I am sure. Some idiots will remain filling their sites with Flash crap and other mindless additions. But many people will turn to the basic idea of the Semantic Web once they discover the power of the new Safari and tools like NetNewsWire that will for sure make good use of the new technologies featured (i.e. the Atom API).
So what‘s missing? The new Finder. Oh, hadn‘t we had already a couple of new Finders recently? Yes. Apple showed off a new Finder almost every 18 months. But it never catched up with Moore‘s Law and is still the stinky little bastard it was on System 7. What‘s wrong? The Finder is slow. The Finder is NOT multi-threaded (although this feature was pretended to be there every now and then). The Finder is NOT stable (it crashed daily on my system). We need a new Finder. It might look the same, but it desperately needs an internal restructuring to make working with the beast a pleasure. Also, the Finder needs transparent support for archives (I wanna see the contents without unpacking them) and nobody on this planet understands why WebDAV can‘t run over HTTPS. This is a top-rank security problem! So Apple: speed it up and make HTTPS ubiquitous in OS X, allowing every app to access anything over HTTPS where it already can in HTTP.
This brings me to another point of praise: although no screen shots or demos have been published so far, the description of Mac OS X Server Tiger promises graphical tools for certificate creation and management. This is so desperately missed in this world today I would pay for the tools now. Hopefully, this will allow universal deployment of encrypted services throughout the workflow.