SAP doesn't get it

Shai Agassi of SAP said:

"But if you look at the most innovative desktop today, Microsoft's Vista is
not copying Linux, it is copying Apple."

But, but, but....

Exactly where does he think Apple is getting some of its innovations from? Just look at Darwin and Safari. Both have strong roots in open source innovation. Apple has been able cherry-pick from successful open-source projects and continue innovating on top of them.

Why do so many pointy heads not realize that Linux isn't the only open source project? Ergh. Open source is now extremely pervasive. And, that's because it continually innovates in ways that commercial products can't.

[/apache] permanent link

A life is measured by both what we remember and pass on.

My friend and fellow Subversion developer Ben Collins-Sussman wrote a touching eulogy for his father.

I am reminded of the Jewish honorific: Zai-kher tzaddik livrakha, which translates into "May the memory of the righteous be for a blessing."

While I certainly have zero insights into Ben's thought processes, but being aware of the baby on the way (much nachas!), I also remember the Jewish customs of naming children after recently deceased relatives - usually not with exact, but similar names.

In my own case, I'm named after my great-uncle J. and my great-grandmother (or was it great-aunt? Doh!) R. who both died shortly before I was born. It's just another way we can choose to honor their memories. And, more importantly, it's also a way we can tell our children (and, hence, to eternity) and pass along the stories - "let me tell you why you are named Justin..."

[/apache] permanent link

AnandTech benchmarks on Mac OS X concerning Apache...

I know several folks have mentioned the AnandTech benchmarks with respect to MySQL. Read the original AnandTech article. I read it too.

Thinking the Apache results were rather fishy (as I have a dual G5 on my desktop and I've seen better performance than they claim), I sent an email to the author. He promptly replied with his httpd configuration. In the article, he claimed he used Apache 1.3 - however, the config file I received was clearly for 2.0. (Definitions of MPMs, etc, etc.)

I've replied back to him pointing that out; and pointing out that I think flood is a far better performance tester than the dreadful ab. I've not yet received any further response from him.

I've heard unconfirmed reports that httpd 2.x is even faster on Tiger than Panther because 64-bit code is available natively and httpd 2.x runs better on 64-bit platforms than 32-bit ones (due to the time calculations and SSL algorithms).

I'll see about getting some 'real tests' this week on my G5. Who knows - I might confirm them, I might not. We'll see. But, not indicating the difference between 1.3 and 2.0 is wonky.

[/apache] permanent link

mail.apache.org was letting newer viruses through...fixed.

In the category of whomever thought this was a good idea should be shot:

The FreeBSD port maintainers switched the default clam AV database default in a point revision in the ports tree. That's fine as far as that goes.

What isn't fine is when the port silently updated the freshclam.conf but didn't do the same to the clamd.conf file. Therefore, freshclam was updating the database happily and clamd had no clue as it was pointing at the old directory. Grr.

The ports should never ever ever touch a live configuration file. I don't care what they think is good, but that's just dumb behavior. clam not using a single configuration file probably isn't that smart, either. It lets people walk into these types of mistakes.

The offending commit to the clamav port Makefile

Anyhow, fixed manually as of yesterday morning.

[/apache] permanent link

Security through obscurity is bad...

One of the arguments that seemed to have been made in favor of dropping the connection instead of sending a 400 Bad Request is that it provides too much information to the attackers. There are arguments against this that were made by others based on how it makes it harder for a client developer.

Well, you point me at a web server, I'm likely to be able to track down exactly what version of what web server and OS you are running even without any explicit information like what is sent in the Server header or the pointless mis-representation of what server you are using.

The only real defense is to have a secure server to begin with.

[/apache] permanent link