Blogtastic

Safari HTTP version?

Very odd - I've just noticed that Safari seems to be using HTTP 1.0:

xx.xx.xx.xx - - [28/Jan/2003:18:39:38 -0600] "GET /xbench/csi.xhtml?machineTypeID=12 HTTP/1.0" 200 4938 "http://ladd.dyndns.org/xbench/" "Mozilla/5.0 (Macintosh; U; PPC Mac OS X; en-us) AppleWebKit/51 (like Gecko) Safari/51"

No wonder its page load speeds are slower than Chimera with pipelining.

An inordinate fondness for beetles

Amazing beetle photography.

I'm working on an 8x10 print using tons of these beetle images.

Wisconsin, In Winter

The snow was falling. An old man slipped on the ice, breaking vital pieces of cartilage. An overworked young man expelled the last of his fluid reserves in the form of phlegm.

A previously unknown chicken discovers a cure for malaria. Five minutes later, it too slips on the ice, and forgets its pivotal discovery.

I love users

I got an email from a user today about the Xbench website, saying:

"It is lame that your site cannot work with Safari. I guess web programming takes real talent."

Needless to say, this annoyed me to no end, so I looked over the Xbench website, and changed the format of the error message that appears when you're using Safari. Oddly enough, as soon as I inserted a new div for the error message, it began working properly in Safari.

Stupid browser bugs...

But, on the plus side, it's XHTML 1.1 compliant now.

I think I'm dying of diptheria today.

D.C. Rally

Jan has posted pictures from the D.C. Peace Rally.

Cocoon

Optimizing Cocoon today, I realized that much of the website documentation is traitorous lies, unless you've moved on to the developmental 2.1 series.

So I ditched 2.0.4, and moved the xbench server code over. So far, it seems peppier, and having a working XSLTC is rather nice. Also the profiler is somewhat useful, although doesn't seem to be able to shed any light on the (rather hefty) amount of time spent processing the sitemap, and other surrounding tasks.

I've benchmarked Tomcat, which is delivering JSPs in under 10 ms, so the problem seems to lie within Cocoon. The profiler gives me stuff like so:

<result time="1340" index="0">
<component index="0" role="jsp" source="compareindex.jsp" time="2" />
<component index="1" role="xslt" source="stylesheets/compareindex2xhtml.xsl" time="51" />
<component index="2" role="cinclude" time="78" />
<component index="3" role="xslt" source="stylesheets/removens.xsl" time="5" />
<component index="4" role="xhtml" time="31" />
</result>

What's odd is that the result time is 1540 milliseconds, yet the sum of the individual steps is only 167 milliseconds. So apparently cocoon is doing lots of other things not measured by the profiler. Also, cinclude is relatively slow in its non-caching implementation. Switching to the cached implementation appear to have roughly doubled the cinclude performance.

Peace

Watched the C-SPAN coverage of the D.C. peace rally today, which my little brother is attending. Looked like a good turnout. Some of the speakers were a bit... interesting. Rhyming "Tony Blair" with "warfare" was a stroke of inspiration, to be sure.

Froggy

Ahh, bug fixes today for Xbench.

Strangely, I'm now thinking about Docktastic.. Perhaps it needs its own community site, with floating image tiles. Yess... Yess...

Jesse and Jan have supplied me with my late Christmas presents. It's a second Christmas, but without the jolly elves!

Xbench 1.0b5

Just released a new Xbench this morning, with the super-spiffy Results Comparison Website.

Had to do some last minute server fixup, and have been adding lots of new features. Naturally everybody's making up their own special names for the kind of Mac they have. Things like "Super-Duper Case-Modded G4/with Mirrors, EXTREME!!". *sigh*, time to revise the UI.

Cartoon Brick Wall

Still reading Blue Latitudes, a gift from Naomi. I find the comparison between New Zealand's Maori and Australia's Aborigines to be quite illuminating. Evidently the Maori were considerably more fierce, even eating some of Captain Cook's crew.

Evidently a thirst for lipids, such as those provided by human fat, is more common in places without many large land animals. So watch your vegan friends - in close quarters, they may suddenly snap, turning to cannibalism, due to insufficient bodily reserves of fat.

In other news, it looks like L337.com has accumulated $43 in revenue over the christmas season. Free money is always nice.

Code, code, everywhere, and not a drop to drink

I seem to have very much more free time lately, now that I'm not encumbered by school and work. Therefore, I have started this thrilling weblog. This MovableType thingy seems reasonably nice, although I think I will eventually write a Cocoon-based weblogger.

The sunset was rather nice the other day. My Elph's color reproduction was not so nice:

My latent lighting lust was re-ignited the other day, on recovery of my Sunday New York Times from my thieving neighbors. Within, I found this:

Today, I'm continuing work on Xbench, with the community website being the main focus.

The HTML design is still very primitive, but a lot of the core features are working. The tricky bit is to get a meaningful display of the different benchmark sections, and compare them between two disparate systems. (Today - how to compare a dual processor system with a single processor system)

I've now ditched eXist for a much simpler MySQL-based solution. I'm still using Cocoon, inside Tomcat, which is quite nice, despite their insistence on dumping compile errors into a different log file for arbitrarily different types of compile errors.

I was hoping for Xpath support in the database, and still hope to get a working XML database solution eventually - I just wasn't happy with eXist's indexes corrupting themselves after every document insertion.

Hello, Fiendish Bloggers

Welcome to Ladd's tin-plate-weblog contraption.