July 2003

Interesting. Ray Ozzie wants email-to-RSS. Me, I've been thinking for a while about the opposite. I do not provide any means for others to comment on what I write (except for sending me an email). I also don't visit my own blogs that often and would't want having to either. Comments would force me to do that. That is, if I can't make the comments come to me, which is where Comments-To-Email comes into play. I read email all the time and would love to read (and answer) comments made on my blog directly in my favorite mail client (which makes me wonder, is there perhaps such a beast available today in, say, Movable Type?).

As I generate my blogs locally on my laptop and push them to the server there's not much opportunity for dynamic content. My thought was to write a set of programs that would take care of this; one program handles submitted comments and sends them to a special address that I'm in control of; delivered comments will end up in my special comment mailbox; other programs can query the comments database and the result can be included in a web page.

I usually do not link to images here but today I'll make an exception because of what Tim Bray writes.

The message is this: If you were looking at this in any browser but Microsoft Internet Explorer, it would look and run better and faster.

It struck me that in a house full of Macs, downloading each update for every Mac isn't very efficient. What if there was a Software Update Proxy service?

1. Client machine C1 selects machine SUS as its Software Update proxy.
2. C1 checks for updates and what do you know, there's a new update available and C1 starts the download.
3. Acting as a proxy for the download request SUS stores a local copy of the update package.
4. Client machine C2 also has SUS as its Software Update proxy. It checks for new updates and what do you know, there's a new update available and C2 starts the download.
5. SUS sees that it has already downloaded (or is in the process of downloading) the update that C2 just requested. It verifies the signature of the local copy with Apple's download servers. If verification turns out ok the local copy is handed to C2.
6. The same procedure is repeated for all Macs in the machine park. Great speedup for most of the Macs, and a great bandwidth saver.

I can see at least two common scenarios:

1. One machine plays the role of a dedicated SU service server. All other Macs in the house selects this server as their Software Update proxy.
2. Every machine could act as a proxy on its own, and announce this through rendezvous. Excellent for a small home network for example, where you might not have a dedicated server.

Just a thought anyway. I filed a feature request at Apple's website.

When you choose create Java source files from an entity in EOModeler you usually get a Template Generation panel in your face informing you that a Class template for [entity name here] already exists. You then have the option to Overwrite, Save as, or Merge. The merge option launches FileMerge which makes it easy to integrate a model change with your custom data.

In an WO5 project however, you can put your files wherever you like. For example, in the JavaClient application I'm writing right now I've put server-side classes in a folder named Server and the client-side classes in a folder named—not surprisingly—Client. When I hit the generate source files menu item in EOModeler I first get to choose where to put the resulting files. I guess the panel is a standard AppKit NSSavePanel, and what happens is that when selecting one of my sub folders where I keep the server or client side source files I get a standard (?) alert panel giving me the option to either Cancel or Replace. Where did Merge go?

I could of course move the source files back up to the project folder, but then I'd have to rename one set of the files as the only thing telling them apart today is their package; the server-side classes have the same file names as their client-side counterparts. Very inconvenient.

Gamasutra, Extreme Game Development: Right on Time, Every Time (requires registration): Extreme Game Development (XGD) is an agile game production method based on the popular software development method called Extreme Programming (XP). XP is not simply another fancy development method; it focuses instead on the one truly crucial issue: creating a working product.

I'm not interested in buying Viagra. I am not interested in buying Viagra. I am N.O.T I.N.T.E.R.E.S.T.E.D in buying Viagra. STOP SENDING ME E-MAILS ABOUT VIAGRA! Please?

Man, do I hate spam...

You better think about the question—and think hard—before answering it: Fallacies of Presumtion.

I stumbled upon this page on the CocoaDev wiki and decided to try and incorporate it in ObjcUnit. It seems to work very well and I think we'll release a new version of the framework soon. Hopefully this addition will address what many—including myself—feel is a shortcoming in ObjcUnit, namely that you can't see the line number of failure. Yay!

The MTA qmail has a nice feature. Deliveries to local accounts are handled by dot qmail files in the user's home directory. The default delivery file is simply called .qmail and contains instructions of how and where to deliver incoming messages. In the default installation of qmail a user also has control of the address name space below his or her username. That is, the user Joe is in control not only of the address joe@example.com but also every address starting with joe-, for example joe-banking@example.com, joe-programming@example.com etc. To set up these addresses Joe only has to create the files .qmail-banking and .qmail-programming and fill them with delivery instructions.

I have a few of these sub-addresses on my personal mail server. However, until this morning I didn't know how to make Mac OS X's mail client Mail recognize these sub-addresses without adding a new account, which I did not want to do (I actually sent a feature request to Apple some time ago regarding this). It turns out Mail has (probably) handled this for a long time. The obvious solution (thank you Urban), which I never tried out, was simply to add another address to the Email address field in the preferences for the account. The address entered in preferences is used by Mail as the envelope sender when delivering mail to an SMTP server (the MAIL FROM command).

From now on, when I compose a new message in Mail the account pop up button is populated with my sub-addresses and the message will originate from that real address. No need to set Reply-To manually, no need to use mutt on the server anymore. Wonderful!