WombatNation

VoiceXML Review used to regularly publish interesting articles and news updates related to VoiceXML, but had fallen somewhat silent over the last six months or so. A lot has been going on with respect to VXML 2.1 and 3.0, so I’ve been hoping to see some more activity on the site. It’s good to see that someone from the VoiceXML Forum is updating it again.

One of the articles at first looked like a totally unrelated article on internship programs at IBM. However, the two interns who wrote about their experiences both worked on speech technologies with members of the Pervasive Computing group at IBM.

The team in Austin used the Opera browser and XHTML+Voice to create a multimodal application for finding info on movies playing at local cinemas. I remain quite skeptical about the value of multimodal applications on PDAs (at least with the PDA technologies of the next 1-2 years), but it’s still good to see people making progress with standards-based approaches like X+V for building multimodal apps.

I’m back from vacation, but am unfortunately one of the people who has suffered due to the current troubles at US Airways. My outbound flight from San Francisco to Orlando (I went on a cruise, not to the Disney greater metropolitan theme park) was operated by United, codeshared with US Air. Unfortunately for me, the return flight was operated by US Air.

Our flight to Charlotte on Xmas afternoon was delayed almost an hour. We would’ve had only fifteen minutes to catch our flight to SFO, but, of course, that flight was delayed, too. Many US Air flights were cancelled, so a two hour delay maybe wasn’t so bad.

The reason for many of the delays and cancellations was that a lot of flight attendants and baggage handlers called in sick on Xmas day. One of the union leaders said it was not a union action, but that he thought it demonstrated bad planning by management. Uh, you tipped your hand with that lame explanation, buddy.

The sickout wasn’t the cause for our flight’s delay, though. Due to FAA rules, they can’t board passengers on a plane until the entire crew is present. We were missing one flight attendant, but not because of “sickness”. Earlier in the day, she had been riding in a van to an airport in the midwest. Her van hit a deer and turned over. Fortunately, she wasn’t hurt, but the delay caused her to miss her flight.

To the credit of the employees who did show up for work, every US Air employee I dealt with was friendly, helpful, and very professional. While I obviously don’t know the inside story of employee-corporate relations at US Air, I hope that all the flight attendants and baggage handlers who lied about being sick received a particularly nasty lump of coal in their Xmas stockings.

While we finally got to San Francisco at around 11 pm, our luggage did not. Eighteen hours later, I still don’t have it. However, according to US Air’s baggage locator automated speech app, it has been picked up by a delivery company and should be dropped off in the next couple of hours.

The speech app was pretty good. Not great, but good. At least it was a lot better than the US Air website. The website just had a form that you could submit. Six hours later and I still have not received a response to the request I submitted. The speech app gave me immediate feedback on the status of my bags.

The biggest problems with their speech app were:

• Too much text-to-speech. The worst case was when “got it” was TTS’d. Also, numbers sounded like they were either TTS’d or were poorly recorded.
• Did not recognize the “lost baggage” file reference number. It failed on two tries to understand the six-letter code, even though I spoke it very clearly and in a way that I have had very good success at work with our speech applications. Their app had to fall back on having me read one of the six-digit baggage claim numbers.

In the end, though, the speech app got the job done quickly and effectively. The web app was time consuming and, ultimately, useless. Speech apps 1 – web apps 0.

I was hoping that Python 2.4 would make MoinMoin a little quicker, and a completely unscientific survey (surveys conducted after midnight are rarely scientific) confirms this. With Python 2.3, opening pages for editing and saving them almost always took at least five seconds per action. While it’s still sometimes slow, the actions will often take only about 2 or 3 seconds with Python 2.4. MoinMoin still feels too sludgey, especially in comparison to JSPWiki which I use at work, but at least it’s now tolerable.

Next up is the MoinMoin 1.3 upgrade. Then, if I have time to follow up on Steve Jenson’s suggestion, I’m going to see if I can use Psyco to speed up the execution of MoinMoin’s Python code. Psyco 1.3 already supports Python 2.4.

Note to Self for Future

The following steps are for installing Python 2.4 on a shared server. After ssh’ing to the server:

$ wget http://python.org/ftp/python/2.4/Python-2.4.tgz
$ cd Python-2.4
$ ./configure --prefix=$HOME/lib
$ make && make install
$ ln -s $HOME/lib/bin/python $HOME/bin/python/bin/python

Credits: Thanks to Uche Ogbuji for publishing IRC transcript excerpts that reminded of the steps I had forgotten, though I did use my somewhat stranger directory structure for the symbolic link.

Nicholas decided to be the Thanksgiving surprise. However, he easily outweighed the twelve pound turkey we had purchased, so it was pretty easy to tell that a switch had been made. That and all the black fur poking through the holes in the box.

I just noticed that Python 2.4 was released at the end of November. Fedora Core 3 RPMs are already available from the Python.org site. It’s supposed to have better performance than 2.3, so I may have to try it on my website to see if it will speed up my MoinMoin wiki, which is frustratingly slow.

Of course, this prompted me to check out the Moin Moin site, and I was pleased to discover that version 1.3 was released last week. New features that will cause me to upgrade soon:

1. An antispam feature that “compares the page content against known spam patterns” when saving pages. A blacklist feature is also included.
2. Enhanced search features, including regular expressions, case sensitive and title search
3. Underlay directory to keep system and help pages separate. This will make later upgrades much easier.
4. The new modern theme is much nicer than the old default theme

My website is on a shared server and I’m running MoinMoin on Apache via CGI. If my site were on a dedicated server, I’m sure I could get much better performance by running it on Twisted or on Apache with mod_python. I’m pretty sure my webhost doesn’t have mod_python installed. I’ve got to figure out some way to find out why it is so slow, though.

I’m very happy about the anti-spam feature. Until now, I’ve not posted any links to my wiki, because I’ve been afraid that spammers will find it.

Okay, it’s clear that I’ll be upgrading both Python and MoinMoin on my website soon. If I can’t figure out a way to speed up MoinMoin, though, I will likely give MediaWiki a try.

If you live in the Western United States, as of today you can get a free credit report from each of TransUnion, Experian, and Equifax. You can do this once per year. Check out AnnualCreditReport.com to see when you get your chance if you don’t live out west.

Update 12/9: I corrected the above URLs. Thanks for pointing out my mistake, Richard!

I was able to view my reports from TransUnion and Experian instantly online. Unfortunately, I will have to fill out a form and mail it in to Equifax to get my report from them.

You might be wondering, why are the credit bureaus doing this for free? I have no idea if the FTC forced them to make this offer, but they each take the opportunity to market a lot of their other services while you’re signing up for the free credit reports. I’m very pleased that the credit bureaus have made credit reports available for free annually, but they are definitely getting a lot of marketing value from the effort.

If you’re like me and have been a victim of identity theft, you’ll definitely want to take advantage of this offer as soon as possible. While you get a credit report for free if you have been a victim of fraud in the previous 12 months, it’s good to check up your credit report after that time period, especially if you don’t know if the thief was ever caught. The good news for me is that the bad credit card info has been removed from my TransUnion and Experian reports. The bad news is that those reports still contain bad phone numbers, addresses, and names. Some day I’ll get around to adding this info to my Privacy and Information Security page.

Update 12/2/2004 – It turns out this was required under the terms of the Fair and Accurate Credit Transactions (FACT) Act that was passed in 2003.