Some small rants and raves for your reading pleasure on this Sunday :)
- Randy talks about active patching of SqlServer. He likes the idea. Well, in theory it's neat. In practise it's hell. SqlServer hotfixes come most of the time in a simple zip or exe which simply packs out a bunch of files and a readme. You should follow the readme file which simply tells you which files in which directories to backup, and which files in the archive you just unpacked should be copied in which directories. Restart the server process and you're patched. This is cumbersome. I've done this kind of patching for years now and I always wondered why SqlServer is not equipped with a good installer for its patches. Nevertheless, active patching is bad. Imagine a machine in a server centre near a big internet access point with SqlServer and it decides to patch itself. What is backupped? Is that backup also backing up active transactions (current SqlServer backup can't do that)? If the patch fails, is my server brought back into the state it had before the active patcher was started? If I start a patch process myself, I can make sure I've done everything I could to make the patch run smoothly. However an active patcher just runs the patch, does it test the patch on my config? Don't think so. So Microsoft: bad idea.
- Brian talks about PSS and hotfixes and that the patches are free. I never understood what PSS has to do with updates for Microsoft software. We're developers. If a patch is available for a given VS.NET bug, release the darn patch! No... if you are really lucky, Microsoft has put up a KB article telling you there is a patch (but you are not gonna get it unless you call PSS) but for most bugs / patches (like the ones in VS.NET), no patch nor KB article is made available. Customers are then left in the dark. Also, PSS isn't free for all people, so in some circumstances you have to pay for a patch for software you already payed for. I'm sure some financial genius has cooked that up, but it's not how a company should treat its customers nor should customers (that's us, dear reader) be happy with how Microsoft treats us. "It's not tested well enough", and other 'excuses' are mentioned why a patch isn't released to the public. I can tell you, Microsoft, that's BS. Remember the de-serialization bug in .NET 1.0 for structs in array objects? That was patched in the fall of 2002. We had to wait till April 2003 for the fix. There is no excuse for that: developers who write software for platforms you sell, Microsoft, are the key to success of those platforms. If you mistreat those developers, you will loose the platforms. As simple as that.
- Yesterday I moved to an AMD XP 2600+ with 1GB PC2700 (kingston CL2.5) and Epox Mobo. I simply did a heart-lung transplantation on my old P3-933 machine with an i815 mobo, did a repair install of Windows XP, re-installed the service pack and the hotfixes and I was back in business (ok, had to re-activate XP by phone, typing in 80 or so digits via the phone... why so many digits?). Really nice feature (that repair install feature of XP, not the re-activate by phone ;))
- Did I already mention that typing documentation is boring? It is boring :)
That's it for now :D