Archive for the 'Research' Category

A quick review of MySQL Workbench

Monday, July 14th, 2008

I recently purchased MySQL Workbench Standard Edition to help me design databases and to reverse engineer the databases of open source projects that I build on and support. I figured insight into the open source projects workings would allow me to extend and support them better.

So far I have only reversed engineered two open source ecommerce projects: Zen Cart and Magento Commerce. Workbench did a great job of identifying the tables, and in the case of Magento building the relations between tables. The layout or autolayout of the results left plenty of room for improvement. I tried to make Magento DB layout better by moving tables around, but Workbench could keep up with the task. When moved, the tables would disappear and then reappear when dropped. If your lucky, the table would be in the neighborhood of where you wanted it. If your not lucky, it would stay put.

All in all, Workbench seems to be coming along but it isn’t ready for serious design work. I look forward to the day that it is.

UPDATE: MySQL Workbench can crash randomly. Make sure you save often. When it crashes, a dialogue box will appear to offer you the option of saving. Save. Do not click cancel. Upon canceling, Workbench will only proceed to crash making you lose all your unsaved work. It just happened to me.

UPDATE: I found a little problem with the SQL CREATE export. At the very end of the CREATE syntax, the table type is specified. For example:

ENGINE = InnoDB;

Sometimes the line is not terminated with a semicolon (;). If the line in improperly terminated, importing the .sql file on the command line throws an error: ERROR 1064 (42000).

Streaming Flash Video

Wednesday, June 6th, 2007

I don’t do much video stuff, but this post on streaming FLVs via PHP at FlashGuru seems like it could be pretty handy. If you are encoding a video into an FLV, the word on the street is that Sorenson Media’s Squeeze for Flash is the best encoder.

Google Hell

Tuesday, May 1st, 2007

Google maintains a supplemental index (known as Google Hell in web marketing circles) for sites that are considered spam, invalid, too out of date, etc. Being in the supplemental index means that your results won’t show in the first few pages of Google search results. For businesses relying on search engines to drive traffic to their sites, this can have the practical effect of closing shop for a few days a week. There is a lot of lost profit from the reduction in organic search results.

The key for web marketers is to be very careful how you market your site. Using a company that relies on ubiquitous linking strategy could be the death nail that kills your profits. Be very careful and monitor your site stats aggressively.

Solutions for IE’s PNG Problems

Saturday, April 7th, 2007

All versions of Internet Explorer prior to 7 cannot correctly display the transparent PNG-24 images. In place of the transparent pixels are light gray pixels. The solution that I like the most is called the IE PNG Alpha Fix. It uses a .htc file that corrects the behavior of IE using javascript. The fix also requires the a CSS hook:

* { behavior: url(iepngfix.htc); }

I have used this solution on a couple of projects now, it works well and doesn’t require gratuitous HTML hackery.

Fitts’ Law and A-Z Listings

Thursday, February 22nd, 2007
Snapshot of Boston University A-Z navigationBoston University’s A-Z Listing

A common navigation element on university web sites is the A-Z list. In general, the lists work as a pseudo-site map. They are effective in getting people to common “tags” quickly.

A-Z listings by their very nature have small hit areas; each letter is linked to a distinct page. These small hit areas are trouble for physically impaired people or people using imprecise pointing devices. Fitz law states that there is a correlation between the distance from the final target and its size to the accuracy of completing the hitting task. In other words, a small link that is far away from the mouse pointer will be harder to click on than a big link that is close to the mouse pointer. A-Z listing are great stopgap solutions for findability, but trouble for usability.

There are a couple of solutions to this problem, but they are not used very often. Below are the A-Z listings for Cornell, Berkeley, and KU. Of these, KU’s is the least ideal solution. Its hit area is small and the spacing between the letters (hit areas) is small. The other two still have small hit area’s but have more spacing between the letters. The spacing should reduce the number of missed clicks.

The first solution to the problem is to increase the hit area for each letter. Larger hit area The hit area does not need to be indicated upon first impression, though that would be nice. It does need to have a indicated that area when the mouse pointer is hovering the letter. This small visual cue will confirm with the user that they are indeed about to click on the right letter.

The other solution is to group letters together. Boston University does this. The grouping of letters automatically increases the hit area, which in turn solves the problem of Fitz Law.

Snapshot of Kansas A-Z navigation
KU’s A-Z Listing
Snapshot of Berkeley A-Z navigation
Berkeley’s A-Z Listing
Snapshot of Cornell A-Z navigation
Cornell’s A-Z Listing

Research library

Saturday, February 17th, 2007

I’m starting a library of papers, articles, and other stuff that I run across and read.