Wow, just wow

August 22, 2006

There’s a very intermittent bug I’ve hit every now and then in batches related to webservers that unexpectedly close connections. It occurs in batches because web site opererators eventually catch on and fix the problem. However, the effect in Fact200 is that it pegs the CPU at 100% and although the app can still run, it, along with the rest of the system, runs very slowly.

Luckily, I came across this blog entry http://nick.typepad.com/blog/2006/06/microsoft_pleas.html

And microsofts eventual response:

http://blogs.msdn.com/wndp/archive/2006/08/14/a_chunked_encoding_story.aspx

Next time my boss tells me to fix a bug, I’ll show him how the folks at Microsoft do it :). Lucky thing he blogged about it and the folks at MS came across it or it might have persisted through Windows Vista! Actually, it hasnt been verified that its fixed yet…..

Edit:

Just to clarify the Wowness: The KB article written by MS (so it’s an acknowledged bug) on the issue is dated Oct 2, 2003 and its stamped as rev 3.0. One would surmise that the discovery date was sooner, but taking the rev 3.0 KB article date would mean almost 3 years for them to fix a bug that pegs the CPU at 100% until you kill the process.

A funny and somewhat accurate depiction of communications in the software development environement.
http://i3.photobucket.com/albums/y64/tmptmp/software_development.jpg

What a coincidence.
Techdirt: Nice Patent… For Me to Poop On!

Having read of various patents (MS’s auto chat matching, and a few on web previews off the top of my head) that seem a little on the obvious side this sounds like a pretty good idea. Of course it could be difficult to sort out malicious intent (from competitors) from true criticisms. It might result in politicization of the whole process
Slashdot | Patent Reviews Via Wiki

1.0b10

August 15, 2006

Another two weeks and another release. As noted in a past entry, this version contains the new tiled view which further evolves the preview features of Fact200. It turns web surfing into something akin to photo album viewing. It works well if you’re visually inclined and makes scanning many pages a breeze.

Thumbnail support was also added to the regular list views, but because these are much smaller (and thus less useful), the default configuration has them switched off.  Both of these features can be adjusted through the preferences dialog under the previews tab.

An experimental feature that’s been added in this version (but poorly documented as usual….) is preloading of pages in the minibrowser when hover previewing. Preloading was already being done when viewing pages in slideshow mode.  It helps minimize the delay when switching from the image based preview to the minibrowser. Because there is potential for a small delay, the default for this option is off. If you find that you’re switching into the minibrowser very often, you might want to test this option out.

To turn it on you’ll need to muck around in the registry: set the HKEY_CURRENT_USER\Software\Excelsis\Fact200\preloadMinibrowser value to 1

Google sees privacy threats

Privacy has taken center stage again with this whole AOL debacle where they accidentally released search data that could be traced back to individuals. While it’s probably true that some firms are better than others at preventing these sort of leaks, I think one of the more important factors is that this information is being collected at all.

It’s not really surprising that this sort of thing is going on. Most marketers would probably sacrifice their first born to get a peak at the browsing behaviors of their target markets. There’s the obvious direct marketing route whereby a user whose been searching for ‘diaper rashes’, ‘milk formula’, or ‘child seats’ could be bombarded by ads for baby products. But there’s also the ability to mine the data further to discern patterns and trends like age groups, locations, brands, environmental correctness, political affiliations, income group, etc. Collecting this sort of data reliably over time can really only happen using a unique id (i.e. cookies) or login.

For the most part, this data gets tied to an anonymous entity or, in the case of a login, whatever information you choose to give. However, things start to get a little closer when you start searching for personally identifiable items. Names of friends, family, co-workers, or the amazingly common vanity search can quickly narrow the identity of the anonymous entity. Ever type in your address? You probably have if you’ve ever gotten directions from a mapping service.

But does it matter? Ultimately it really just depends on how much you value your privacy; some people live life with the windows open and the lights on while others prefer a little more discretion. But because these are businesses we’re talking about, you can be sure that this information will be put to commercial use, whether it be for advertising on behalf of others or for gaining competitive advantages for the self. In the end, privacy will always be a potential problem with server based systems that can identify you with either an IP address, a cookie, or a login.

Fact200, being a thin client application, can help alleviate these concerns. Of course, when operating as a thin client, it must connect to a server and is thus susceptible to some tracking. However, Fact200 takes great pains to connect to servers without using persistent cookies. This prevents tracking through most means except IP addresses. If you use DSL or Cable, chances are good that you have an IP that is both shared and dynamic thus making identification difficult. You can further get around this via an onion routing network like Tor.

Conversely, with servers and integrated services, there’s almost always a path to trace: e-mail provides a good consistent ID to start with, search provides general traits, and mapping provides location.

Fact200 also maintains its own independent data library that resides on your local computer. Searches through this library do not go over the Internet and are thus untraceable. Even though the source of all data is from the Internet, correlating any sort of search behavior would be difficult. For example, one might do a broad search for ‘porn’ which, without cookies, would already be hard to trace. But then one might then do a library only search for ‘midget’ thus obscuring the original intended search from the Internet.
Data sources like news sites or social bookmarking sites can also be very broad making it difficult to ascertain any distinguishing individual characteristics. Ultimately, the more often you search within the library, the closer you will be to absolute privacy.

Zachary’s San Ramon

August 9, 2006

With the changes necessary to correctly integrate with del.icio.us, integration with blogging sites like WordPress is now trivial. Right now, it just sends the link and nothing more, but a custom blog type of dialog will help flesh out the content more and perhaps even an option to publish a link outright will be coming; right now it only gets saved.

Zachary’s San Ramon

1b9 is out, featuring realtime search within results (like Firefox) along with HTML export support of result lists and a few other things I’ve already forgotten because…

No sooner than have these neat little features been released that I began work on another cool feature that I’ve been thinking about for a long time. While the previous features carried Fact200 textual processing a bit further, this new one focuses on previews and image processing.Tiling Sample