Blog - The Geek under my Skin

As I am switching my calendar and contact from a Microsoft Exchange server to my Google Account, I needed to make the configuration change on my T-Mobile Dash as well. Google offers an ActiveSync server to sync Windows Mobile devices with your Google Calendar (m.google.com). You can find more details about that here.

The issue I faced though was that the ActiveSync Menu “Configure Server” was letting me change the server, the account password, but the username and domain where not editable. After a few minutes, I finally figured out the trick to make the switch to a different ActiveSync server: you must delete the ActiveSync configuration first.

From the ActiveSync screen:

1. Menu
2. Options
3. Put the selection on “Microsoft Exchange”
4. Menu
5. 1 Delete
6. Yes to the alert

and your previous Exchange server is now gone. Hit “Done”, then “Menu” and finally “Add Server Source”, fill in the server and user information and you will be sync’ed up with your new server in no time.

The Nokia N800 Internet tablet with OS2008 supports Adobe Flash 9 for playing video and running small Flash apps. But when I tried using websites which record video using Flash, I was surprised to see that no webcam was found. The N800 clearly has a webcam, which is supported by the provide VoiP client.

After some Googling, I found the answers to my questions: Adobe Flash 9 on Linux only interface with the now deprecated Video4linux API version 1. The Linux kernels for the past couple of years have stopped compiling the Video4linux 1 support some time back, in favor to the new version of the API, Video4Linux 2.

While searching on this topic, I found this small utility which basically recreates a v4l1 virtual device based on the v4l2 physical device: Flashcam. A .deb for the Nokia N800 is available ready to install from here: www.zdez.org/flashcam-0.10.deb.

Installation was as simple as can be from the tablet, and in no time, I was able to record videos on Flash websites. The only remaining problem is with audio, which I did not figure yet how to address. This will be an Internet search project for another day.

Google has been running developer events in several cities in Europe in these past couple of months, starting with London, and following with Paris, Munich, Milan. Today was the Prague, Czech Republic event. I was unable to attend previous sessions, and as I am in the Berlin area, Prague is only a 3.5 hour drive away so it was an easy decision to go attend this day of technical sessions on Google technologies.

The day was organized in a fashion very similar to the Google IO event in San Francisco, but in a much smaller scale.

The event was very well organized. The registration process was painless with no waiting. I do not speak or understand Czech but I still had a great choice of sessions in English language, which was great. I have to mention though that although they insisted on us returning feedback forms, they did not provide them in English, so I couldn’t fill mine out. We were provided with a cool grey T-shirt showing the icons of Gears, AppEngine, Android and Maps. The Clarion Hotel conference level had open Wifi service allowing the attendees to stay connected.

We were greeted with a breakfast before the keynote started.

The content of the keynote was pretty much similar to what we had on the first day keynote at IO. Nelson Mattos, VP for the EMEA region, presented in English, but before he started, a lady from local Marketing management kicked off the event in Czech. Following her, a local Engineering Manager also said a few things in Czech before introducing Nelson. I do not understand Czech, therefore I have no idea of what was said in the first 10 minutes of the keynote.

Nelson’s presentation covered the 3 current main focus for Google:

1. Making the browser more powerful (Chrome, Gears)
2. Making the cloud more accessible (Data and AJAX APIs, App Engine)
3. Keeping the connectivity pervasive (Android)

Several other Googlers handled technical parts of the presentation, starting with Mads Sig Ager, a software engineer from the Chrome development team, who covered the Chrome browser presentation and demo, focusing on its features and stability (one process per tab, the built in task manager).

Nelson covered briefly Gears new features, including the geoLocation, openFiles(), and onProgress() handler, before switching to the cloud connectivity, where he made a few statements regarding the Data API for read/write access and the AJAX API for read only services. Vladimir, a local Czech Google engineer performed a few demos, all commented only in Czech. I noticed he showed some stuff using the Search API, demoed some mashups including a race sync’ed on Youtube and Map. His closing demo was labeled YouTube HD, where he had a 2×2 matrix of YouTube players, showing 4 sync’ed videos.

Nelson continued with a quick App Engine update, reminding that this was still pre-release, but that the free quota was allowing apps with up to 5M page views per month. He mentioned that serving https was now possible on App Engine. His then moved to the last focus area with a quick status on Android, and the fact that the source code was now available.

The closing topics of the keynotes were GWT, presented by Andrew Bowers, and Open Social by Nelson. Both these topics were spanning pretty much all 3 focus areas.

The rest of the day was segmented in five 1-hour sessions. There was a good diversity of topics, most in English but about 1/4 in Czech. There were as well workshops on App Engine and Maps API.

I attended an App Engine presentation by Kevin Gibbs, the tech lead of the App Engine team. Kevin shared a bullet item road map of what they are planning to add to the platform between now and end of Q1′09. One of the item was the support for another language (only Python is supported now), which prompted several questions to find out more about what this meant, what language would be added. The answers were not satisfactory though, as Kevin pretended they still do not know which language will come next…

My next session was on Maps and Earth API, which covers a lot of content. Jean-Laurent Wotton was entertaining, making the presentation fun to listen to. But this was a 101 type session, and I wish some topics would have been covered a bit further.

Next, I attended a GWT technical session, covering Javascript Native Interface, Overlay Types and DOM manipulation. As the session ended early, I jumped to the room where NetLog was introducing their OpenSocial platform. It was noted that Netlog does not allow external links on the applications. Such a hard statement will probably keep many developers away from their platform unless they relax this a bit.

My next stop was at Jean-Laurent Wotton YouTube API presentation. Here again, the talk was fun, illustrated with many demos. A lot of functionalities were covered, but experimentations are required to follow up with what was presented. YouTube is a terrific video hosting and streaming platform, but it seems like the capability to host private, not searchable videos is no longer possible. This as well will need to be researched further.

The last session attended was about how to make App Engine applications scalable, with examples like a counter and a blog engine. This was one of the most technical sessions I attended today, but the presenter was going really fast on some aspects. Some slides with 5-10 lines of codes were only shown for 5 seconds or less. :-)

In the middle of these presentations, we had lunch and snacks available, and the day ended with a dinner and a party where a guitar singer was performing live pop songs. We played a few foosball games, had some drinks and food and talked to other attendees before calling it a day.

This event offered a great opportunity to learn about some of the large portfolio of Google technologies. This was a good complement following my attendance at IO 5 months ago, as I could focus on attending sessions in different technical areas. The size of a local event makes the logistic much easier: all the sessions are within a few 10s steps of each others, the amount of participants is manageable, avoiding long lines to register, move around, getting food, etc.

I am looking forward attending another Google Developer Day.

I use several text editors, because I like to check what is available out there. But for quick source code editing tasks, I enjoy using Notepad2, an open source free piece of software from flos-freeware.ch. This comes as a drop in Exe without any installation or configuration utilities. Windows Vista does not like this much, and have been bugging me with the security popup saying “The publisher could not be verified…” for sometimes now. There is a checkbox on this popup to not be bothered again with this warning, but with Notepad2 copied the Program Files directory, the checkbox had simply no effect.

Some Googling on how to address this did not help much, so after some messing around, I successfully cleared the problem doing the following:

1. Copied the notepad2.exe from the Program Files directory to my Desktop
2. Ran notepad2.exe by clicking on the icon on my Desktop
3. On the warning popup, unchecked the box such as I do not get warned next time
4. Closed Notepad2
5. Verified that clicking on the icon on the Desktop again opened without the warning
6. Copied back the notepad2.exe from the Desktop to the Program File location.

The key was that notepad2.exe had to be outside of the Program Files directory to allow the “do not warn me again” option to works.

Of course, after you do this, you will have to recreate your links to the Exe.

The past 2 days attending Google IO in San Francisco have been extremelly exciting. There were many great speakers and many opportunities to meet with Googlers and other attendees.

Generally speaking, the event had been well organized, but it looks like Google has been a bit overwhelmed with huge success. On Wednesday morning, it took about 45 minutes between the arriving at the Moscone center and being able to get on the conference floor. The following pictures give an idea of what this was about while waiting to get our conference passes:

As you would expect, we were greeted with a good breakfast, and had a little bit of time left to network while waiting for the keynote to start. Another hint that the event was not prepared to get that many people is that the open access Wifi network was getting saturated. Connecting to Gmail from my Nokia N800 was a real struggle. There were access points all over the place mounted on small temporary poles. The T-Mo cellular data network was working ok though.

The Wednesday opening keynote was hosted by Vic Gundotra, a Vice President of Engineering at Google.

Several people were standing in the back. The keynote started 15 minutes late, and not everybody had had the time to register, and were allowed in the keynote without their conference badge… The developer community response to the event has been very strong, and there was a big crowd.

The keynote was well organized, and covered the main areas of focus for the conference:

• Gear, with a demo by MySpace of their new mail search
• Android, with live demo of the phone including the street view app using the phone compass to orient the view
• App Engine, where the addition of mem cache and image manipulation libraries were announced, and the fee schedule was finally made public (still free for less than 5Mpv/month)
• AJAX and GData API, with demos of YouTube, News, Transtate, etc.
• GWT, where the version 1.5 was announced, with Java 1.5 support
• Open Social, with a short summary of what is coming in version 0.8 of the API, mainly RESTful API and Template language. Some audience number were made public: In May 08, Open Social is exposed to 275M users, is backed by 20k developers, and has currently 50M application installed

The day went by very fast, attending as many great sessions as possible, including talks on AJAX, GWT, Android, Open Social, Social Graph API. Several sessions were packed, with people sitting on the floor or standing in the back. It was good to see the speakers well prepared and with great public speaking skills.

We stepped out of the conference after the last session to meet to have some dinner in a few friends and Googlers and discuss about some on going projects. When we returned the the Moscone center around 8pm, the after hours party was well on going. There was lot of food and drinks available everywhere, arcade games, Wii on projectors, air hockey tables, pool tables, and live music.

On the second day, the keynote was hosted by Marissa Mayer. The talk was interesting and entertaining, but was mostly about topics we have heard before in other Google talks, or read about on blogs and such.

The second day went by even faster than the previous one. There were more 201 sessions than on the first day, and the quality of the content and the speakers was still outstanding. Session attended included talks on Design Patterns in Javascript, OpenID and OAuth, Open Social, and GWT. At 4:15pm, the event was over. We stayed around for an improvised meeting with some people, while the Moscone staff was taking down Google IO 2008.

I put together a tutorial on how to write an iGoogle Gadget using
MyVox. MyVox is a free service which gives voice to your applications
or gadgets.

If you go through the whole tutorial, by the end you will have your
own working Gadget. It’s also a great experience for those of you who
are relatively new to MyVox, and trying to figure out the best way to
get started. You can see the tutorial here, and the gadget itself is at this location.

I have another set of gadgets, a voice blog management and client
gadget, using this API here:

1. www.google.com/ig/directory?url=api.myvox.com/demo/igoogle/voi…
2. www.google.com/ig/directory?url=api.myvox.com/demo/igoogle/lis…

For more information on MyVox, check out this site: www.myvox.com/

Google opened their Orkut.com social network site to 3rd party applications in limited markets a couple of weeks ago, starting with about 20 test applications. The test markets are still limited to Estonia (1% of the Orkut user base) and India (10% of the Orkut user base).

LjmSite first Orkut application has been included in the Orkut directory for users in India and Estonia about a week ago. We are application number 31.

This application development started back in late October 2007 when Google organized the first Open Social Hackathon in Mountain View that I attended with a limited group of web app developers. We have been working on improving the application functionalities, and porting it to the multiple revisions of Open Social through the past few months. It is a great pleasure to see the application now live with many users trying it out.

The Friends’ Time application allows you to specify the location of some of your friends and displays the time it is at your selected friend locations on your profile. Friends are grouped by time zones. If a friend install the application and set his/her time, this will set the time shown on your profile as well. This application will be of a great help when you want to give a call to a friend who is oversea, or who is travelling, as you have a one stop location to know what time it is where they are. If you are trying to organize an online gaming party, or a video conference call with friends around the world, the application will be a great help with its built-in meeting planner.

The core of the application is based on our popular World Clocks iGoogle gadget. The production Orkut application is using Open Social version 0.7 to get the list of friends and handle the storage of the settings, and the message bundles for localization of the time and date.

We still have several features planned to enhance the application that we will develop in the coming weeks. If you are not in India or Estonia but you would like to check out the application, you can either set your Orkut account to pretend you are in one of these countries, or just patiently wait for Google to deploy Open Social in other markets in the next few weeks.

It sounds like there is another Google Accounts outage ongoing. This is obviously not affecting all Google users, but the Google groups are starting to relay the panic of many locked out users. Once you start using Google services, you depend heavily Gmail, Google Calendar, Google Talk, Google Groups, etc. for your communications and information access. Any downtimes of more than a few minutes becomes a significant pain and distraction.

Two hours ago, I noticed I could not log in my Gmail account anymore. Error was saying that this was a temporary error and to retry in a few minutes. This did not sound too bad.

But after an hour, and noticing that all the Google services using the Google Account became inaccessible, I started digging for information. I filed a problem report with the Google “contact us” link from here.

I found a fresh thread in the Gmail Help Google Group about the exact problem

groups.google.com/group/Gmail-Help-issues-logging-in/topics

I decided to add my comment to let people know how to report the issue, but… rats… with my Google Account being down, I could not add myself to the group to be allowed to post a message… As I have a 2nd account for testing, and it was not affected by the outage, so I was able to post in this thread anyway.

Update - 3pm CST the account is back. The outage lasted a little bit over 2 hours.

The wait is finally over. A couple of weeks ago, I received an email from Bugzilla on maemo.org for a bug I stared some time back. The bug number 1195 was about the missing PAN Bluetooth profile on the N800, which bug had been closed as will not fix some time back. bugs.maemo.org/show_bug.cgi?id=1195

Internet connectivity on the N800 is possible either through Wifi, or through Bluetooth DUN. The DUN profile is supported by Nokia cellphones, but Windows Mobile phones no longer have this Bluetooth profile. Microsoft made the switch from DUN to PAN starting in Windows Mobile 5. My HTC T-Mobile Dash running Windows Mobile 6 had been unable to talk to the Nokia N800 so far, which was limiting greatly the use of the N800 in locations where no public Wifi network was available.

But the good news from this Bugzilla email was that a garage.maemo.org project finally made an installable package available to enable PAN on the N800: garage.maemo.org/projects/maemo-pan

Tonight, I followed the short instruction from the project forum, installed the .deb on the N800, turned Bluetooth on, and enable Internet Sharing on my HTC phone through Bluetooth PAN. I had the surprise to see the N800 reboot twice on the first 2 attempts to establish the connection, but the 3rd try was the good one, and I have been able to browse several websites successfully while being connected to the T-Mobile Edge network.

The N800 will surely travel more often in my pocket now.

Just a quick note about a annoyance I recently experienced with Internet Explorer 7 and the prompt() Javascript function.

Even though doing Javascript prompt() is a poor practice for creating web UI, this is on the other hand a quick and easy way to handle user input without having to implement a form, input, submit on your HTML/CSS UI. For in-house quick and dirty tools, this is something I have happily used in the past.

Well, a few days ago, an Internet Explorer 7 user told me that he was getting a security warning from the browser when using the web app instead of getting the window to enter the data. Sure enough, the web app was still working great on Firefox 2. Starting with Internet Explorer 7, Microsoft disabled the Javascript prompt() function for ’security reason’. You can go back to your Internet settings to revert this security measure, but we should not expect our users to do this. Javascript alert() is still working fine though.

I think we should be thankful to Microsoft, as they are clearly helping us avoiding the quick and dirty solutions. Well,… probably not. 😜

Good news today for the Google Gadget Developers monetizing their work with AdSense. A Google Support Engineer finally clarified the situtation regarding AdSense in gadgets. You can read the post on Google Groups here:

groups.google.com/group/Google-Gadgets-API/browse_thread/thread/050de3cd5fb5d475/b74fe78914f6856b#b74fe78914f6856b

This posts confirms that the lost of relevance of ads in gadgets is the consequence of a technical issue, not a voluntary action to align with a policy change.

If you haven’t followed up the development of this story, the issue started around December 19th, 2007, when we noticed that AdSense ads on iGoogle gadgets stopped being relevant to the gadget content. We noticed as well that the AdSense “Site Diagnostics” reported many blocked URL because of “Robots.txt File” on the *.gmodules.com servers (this is where the Google Gadgets are rendered and served from).

Several developers started engaging with AdSense support, and with Gadget Support Engineers to get the issue resolved. With the holidays season, things started rather slow. On January 10th, 2008, an AdSense support engineer dropped the bomb on us with the following statement:

Please be aware that the Gadget Ads department has changed its policy and it is no longer allowed to use AdSense within Gadgets. The Help Center communication and policy pages are in the process of being updated and I apologize for any inconvenience this may have caused.

This statement caused some confusion, some frustration, and some panic among the Google Gadget developers community. Not much progress was made for a month. But mid February, I was able to get a verbal confirmation that AdSense on iGoogle was allowed, and that there was no plan to change the policy to change that.

Finally, today’s post on the Google Group make an official statement in this direction.

On February 14 and 15, 2008, I came back to Mountain View, CA to visit Google and participate in another Hackathon. My previous visit, late October 2007, for the first Open Social Hackathon, was restricted to an audience of trusted testers and developers, under NDA, therefore I could not blog about this previous event at the time. This new Hackathon was focused on Open Social for the Orkut platform in preparation to the launch of gadgets on Orkut.

It was good to have the opportunity to meet with Googlers working on Orkut and Open Social, from product managers to support engineers. The same event was on-going at the New York Google office, with a video link between both locations. Several of us demoed the gadgets we had in development. I presented on both days the Friends’ Time gadget, which is a port of my iGoogle World Clocks for the Open Social platform. On the second day, my presentation was focused on showing how gadgets for Orkut could be very easily localized using the Google Gadget API Message Bundles.

A topic of importance for Orkut gadgets developers is monetization. Amar, the product manager for Orkut did not have final data to share with us, but he confirmed once again that monetization will be allowed, most likely limited to the canvas view of the gadget. The canvas view is owned by the developer. The profile view is the Orkut user front page, so ads on this view with limited space would affect the user experience. The Orkut team will have a policy covering monetization in early April 2008.

The 2 days of Hackathons allowed me to have some one-on-one or small meetings with a few other Googlers, including a very productive discussion with some people from the iGoogle team to discuss several issues we faced in the past few weeks, and to propose some ideas on how we could avoid them in the future.

Overall, these 2 days on the Google campus have been very productive. They allowed me to spend focus time finishing up my initial Orkut Open Social gadget. This was a good place to network with other developers and share issues and ideas, and most of all, to meet again with Googlers to learn first hand the latest and greatest on Open Social and Orkut.

Below are a few pictures I took on my second day with my cell phone. Enjoy!

A trip to Google always starts with busy traffic on the 101, this time from SFO to Mountain View.

A view of the Googleplex from the distance

The Hackathon was taking place on the Crittenden campus, a group of 4 new buildings

The late October 2007 Open Social Hackathon took place in the 1500 building, but this time, we were located in a larger room in 1400.

Hard working developers, fed with pizza, fresh fruits and mini-cheesecakes.

A short visit on the Googleplex for a meeting

Leaving the Google Crittenden campus at night, to get back to SFO.

I recently starting using a Windows Vista Home Premium machine on my home office network. I have several other Windows machines, mostly XP Home, and we all use a Samba server on a Linux box.

With Windows XP, this has always been working like a charm, but the first few attempts to connect to my Samba user share from the new Vista box was inconsistent. Most of the time, the authentification was failing, but randomly, some attempts ended up being successful.

Having to try sometimes 10 or more times before to login was ok for the first couple of days, but a permanent solution was badly needed. Some Googling quickly highlighted that Vista default security policy uses NTLMV2 authentification, which is only supported on the most recent Samba 3.0.x. My Linux distribution does not provide the required Samba version in the stable depot yet though.

There are several documents which explain how to modify the Local Security Policy on a Windows Vista Business, like this blog post

www.builderau.com.au/blogs/codemonkeybusiness/viewblogpost.htm?p=339270746

But the secpol.msc is no where to be found on Vista Home Premium.

I finally found some forum posts discussing modifying the following key in the Windows registry with regedit:

HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Control\Lsa

Key: LmCompatibilityLevel (DWORD) / Value: 1

On my system, the default value was 3. Changing this back to 1 solved the issue permanently.

Most email accounts receive phishing emails nowadays, hopefully properly identified as such by the email clients, and set aside. Email phishing is becoming just like spam: we get it, filter it, delete it, and do not pay much attention to it anymore, as we know there is very little that can be done until the global Internet email system get revamped significantly.

Today, I was surprised to get a phone call from an automated system reporting some issues with bank transfers and prompting me to go to a website, regionscare.com, to address the problem. As I am not a customer of Regions, this was obvious that this was not a legit call. I had a quick look at the website I was directed to, and sure enough, there was an ok looking page just waiting for an inattentive user to provide his user and password:

After you enter a login and a password, the second page asks you for your ATM card number, expiration date and pin.

A whois on regionscare.com gave an individual name and address, nothing related to Regions bank (this individual is most likely unrelated to the phishing, so will not copy the whois report here).

I have heard and read about phone phishing before, but this is the first time this is reaching my home. Even though we can expect most people to be carefully enough not to fall in this type of trap, adding a phone call in the phishing scheme is surely going to trick more people than just yet another phishing email. With VoIP to PSTN gateways, setting up a system to make these automated calls is fairly trivial and inexpensive and does not required any special call center type infrastructure to contact many phone numbers in a short period of time. Unfortunately, phone phishing is only going to get more common, as will phone spam, and the end users will have to use more techniques to filter out unwanted calls, just as our email clients evolved to filter and delete spam and phishing.

T-Mobile stated a few months back that their Dash owners would receive a free update to Microsoft Windows Mobile Smartphone 6, from the version 5 delivered with the device. I have been watching for the availability of the update from time to time. Today, seeing the update had been available for a couple of days, I took the plunge and upgraded my phone.

I got the update from this page www.t-mobile.com/wmupgrade/ and downloaded the file Dash_Software_Update_1.22.531.4.exe. This is an application that runs on your Windows XP computer from where your smart phone is connected through USB.

The update went fine: no problem, no surprise. In less than 10 minutes, the phone was upgraded and rebooted in Windows Mobile 6.

The only lowlight of the upgrade is that you end up with factory presets. Your email server settings, signature, sounds settings, etc. all get lost. The installation software warns you a few times about that, so you can not say you get caught by surprise, but this is kind of disappointing that the upgrade application does not attempt to keep as much as the smart phone device settings as possible, especially because you get to enter exactly the same settings on WMSP6 as you had to with WMSP5, the exact same forms.

In addition to the upgrade of the phone, you are required to upgrade ActiveSync from the version 4.2 to 4.5 to keep your smart phone and Windows XP talking. You will be reminded by the phone otherwise.

The most noticeable benefit of the upgrade is on the Pocket Office applications. The viewer only versions of the version 5 have been upgraded such as you can now edit documents in Word and Excel. This will be useful to make small edits to documents on the go. I hope the overall stability of the OS will be improved, such as we are not required to reboot our phones weekly as this has been the case with Windows Mobile Smartphone Edition 5.

When you tell colleagues and friends that you are taking a vacation, you most often hear from them the good advice of leaving your laptop and smart phone home. Well, this is most likely a good suggestion only if you are looking forward dealing with over a thousand of emails when you return to the office after being away for a week. In addition to this counter argument, a true computer geek can not live without a daily connection to the Internet.

Anyway, I just came back from a week on a beautiful cruise ship, which took us to the Easter Caribbean and the Bahamas. I, of course, ignored my colleagues and friends advices, and brought my laptop with me, knowing the Caribbean Princess cruise ship had satellite Internet service available.

The Wifi coverage is only available in the middle of the ship on a few levels around the service desk. There is as well an Internet Cafe where you can use computers for a fee if you did not bring your own.

Getting online is a breeze. You are prompted by a captive portal to identify yourself with a login which is based on your stateroom number and your name, and an initial password which is your date of birth:

Once logged in, you can access the Internet at a decent speed with long pings as you can expect with a satellite connection. I measured pings to www.google.com to usually be in the 800-900 ms range.

The main issue though is that Princess states that they don’t support VPNs, and they are not kidding! The technical staff available in the Internet Cafe said this is because they are being a firewall (yes, and what? NAT transversal of VPN protocols works everywhere else nowadays). My couple of attempts to connect to my company IPSec based VPN from Princess private network IP address (a class C on the 10.154.91.0 network with 10.154.91.2 acting as the gateway and DHCP server) each failed.

I was not looking forward reading hundreds of emails through the Webmail on my company public website, especially with the Princess Cruise bill increasing at 75 cents/minute. Luckily, I had my smart phone with me, and decided to give a shot to see if ActiveSync would go through the firewall. And as expected, it did. Therefore, this is how I accessed my email for the week at sea: I plugged my smart phone through USB to my laptop. Microsoft ActiveSync was kicking in, and was then becoming the Internet connection proxy for the smart phone. Once connected to the Internet, I just requested a manual sync from the smart phone and the first 5 KB of each of my new emails were flooding in (I mostly wanted to do email triage, so 5KB was more than enough). I was then able to read and do bare minimum replies from my smart phone whenever/wherever I felt like it, until the next sync through the laptop.

After a few reviews of free and open source weblog engines, and after some experimentation with both Pivot and WordPress, I selected WordPress for its nice and easy to use GUI, and for its backing by a huge community of users.

Pivot was an attractive solution to me as it did not require a SQL back-end, and was storing is data on the web server file system, making backup of the weblog working out of the box with my backup policy already in place. This was as well very well integrated in my web workflow, as I would have been running Pivot on my home web server for the publishing part, and would have been sync’ing the result using sitecopy as I do for my hosted web sites.

But, after playing some with both Pivot and WordPress, my final decision went with WordPress. I can’t really go wrong, as the Import functionality offered in most weblog engines provides almost seemless transition from one engine to the other.

On my development server, I had to create the database for WordPress:

create database WordPress;
GRANT ALL ON WordPress.* to WordPress@localhost
IDENTIFIED BY “secret_password”;

and after that, I was up and running in a couple of minutes.

On Ljmsite.com, my hosted virtual web server at OVH, the MySQL server is on a separate server, and only one single database is available for my hosted account. As I have not been using MySQL on Ljmsite yet, this required me to read some documentation and FAQ from the ISP, but this was still painless to install WordPress there.

Following my first Google Gadget development project, which was pretty much only about putting some XML wrapping around an existing Adobe Flash application, I have started the development of a real Javascript gadget which displays the times and dates of the multiple engineering sites I work with on a regular basis.

The state of the current proof of concept is sufficient for me, and a few collegues, but is of little value to the rest of the world, given that the few time zones supported are non configurable at this point.

Here is a snapshot of the gadget:

The gadget XML is available here on Ljmsite.com.

The next step will be to generate a complete table of the current world time zones from the TZ database (the TZ database contains too much historical data, so only the relevant data must be extracted first). This will take a while to get this done.

After configuring my Google Personalized Home to provide relevant up to date information to me (Slashdot headlines, stock quotes, etc.), I went looking for a gadget to display the time at other engineering sites I am working with. I did not find what I was looking for, so I decided to create my first Google Gadget. My requirements were simple: to be able to display multiple clocks at the correct time of the selected locations, and the gadget should display multiple time zones.

After some quick research, I found that calculating the time at a given location is not an easy task, as some places have day light saving, some don’t, and anyway, the date when day light saving starts or ends is not consistent accross regions or years… I started gathering data on timezones and all, but found a customizable Flash analog clock from WorldTimeServer.

The Flash application takes the country/city as a parameter, so all I had to do was to create a wrapper around this Flash application to make it usable as a Google Gadget. The HTML part was very easy, as a simple table with two Flash analog clocks and a label underneath each was all I needed. I had to process the timezone parameters to format them as proper options for the Google Gadget. VIM helped to get the data properly formatted.

The end result provides me with what I was looking for:

The gadget XML is available here on Ljmsite.com.

I submitted it for inclusion in the Google directory. My next Google Gadget project will be to make digital clocks so that I can squeeze more time zones in the same space, and handle the time zone translation in Javascript, which will remove the need for an external Flash application.

Update: The analog World Clocks gadget is now in the Google Gadget directory.

A new release after over 2 years without published update.

Main changes include:

• Moved the actions to POST forms instead of GET links, which should avoid a crawler to delete/modify the programs
• Added the ‘edit’ functionality (delete + pre-fill input areas)
• Pre-filled the input areas in case of failure to add a program
• Fixed a bug related to v0.5 atd time formating

The theme has been modified to blue colors (CSS changes only).

More information available on the WVR project page.

On my last trip to Hyderabad, India, I flew between Atlanta and Frankfurt, and Frankfurt to Hyderabad on Lufthansa. Out of four 4 flights on Airbus 340, three were offering the Connexion by Boeing Wifi satelite Internet service. I used the free 30 minutes coupon provided by Lufthansa staff to promote the service, and bought some more time by credit card (from $10 for an hour to $27 for 24h, including connecting flights).

The service works very well. I was able to VPN in my office to read/write emails, to access all the web applications like the Wiki for my software projects, eRoom, our Rational ClearQuest defect tracking system, etc. I instant messaged with a collegue travelling on the same flight, located a few seat rows away. I even placed a couple of phone calls through the Wengo SIP service. The latency was a few hundreds of milliseconds long (usually between 500 and 1000 msec), so the phone conversation dynamics is impacted by that, but once we are careful about that, this is usable.

Bottom line, this is technically a terrific service. The only issue I faced was the lack of DC power for theport on coach seats on the Lufthansa A340. My Dell Lattitude D810 with 2 batteries can not go much over 6 hours when Wifi is on.

Eventhough there are Cat5e drops at every corners of my home, a started to feel the need of being able to get mobile with my laptop. I grabbed a Netgear MR814v2 router to use mostly as an access point, and as usual, I cracked it opened to see what was going on in the little box.

I found a Marvell chip is based on an ARM946 core, 4MB of RAM, 1MB of Flash, but I did not find any ASCII string matching a known real time OS. More information and PCB pictures are availlable on this page, formatted to suit the HRI project on Source Forge:

www.ljmsite.com/tech/Netgear_MR814v2/

I have been using this old Linksys router for a few years, but it started acting funny, crashing from time to time. I replaced it with a DLink DI-604 spare router I had, and I opened the Linksys to troubleshoot why it was now having some problems.

In the neat blue and grey box, there is only a Samsung chip based on a ARM7TDI core, and half a meg of RAM, and half a meg of Flash. I was not able to figure out which real time OS was used in the firmware, as I did not find any known ASCII string in there.

More information and a PCB picture are availlable on this page, formatted to suit the HRI project on Source Forge:

www.ljmsite.com/tech/Linksys_BEFSR41/

Uniden, a company mainly known for its wireless telephones, started a product line of home networking product. I grabbed their entry level router which was offered for free after rebate, just for the fun of disassembling and peaking on the internals of this piece of electronic.

I found a Micrel chip is based on an ARM922T core, 8MB of RAM, 1MB of Flash, powered by the Nucleus real time OS from Accelerated Technology Inc, a division from my employer, Mentor Graphics. More information and a PCB picture are availlable on this page, formatted to suit the HRI project on Source Forge:

www.ljmsite.com/tech/Uniden_ENR1504/

Only minor enhancements included in this release.
Added a sort of the at jobs and the cron jobs before display. Cleaned up the Files view (remove % disk space).

More information on the WVR project page.

This is always amazing to see what is really in the box of consumer electronic products. I grabbed a DLink DI-604 rev E router to replace my previous router, but could not resist to the temptation of finding out what was powering this little device.

I found a Marvell chip is based on an ARM9 core, 4MB of RAM, 1MB of Flash, powered by the ThreadX real time OS from Express Logic Inc. More information and a PCB picture are available on this page, formatted to suit the HRI project on Source Forge:

www.ljmsite.com/tech/DLink_DI-604/

Another small release modifying only the back-end driver scripts. Added new recorder scripts to create Low and Medium quality DivX in addition to the mjpeg scripts.

No change to the PHP frontend.

More information available on the WVR project page.

This release does not change the front-end. Only the back-end driver scripts have been modified. The main change is to auto split mjpeg files to keep them under the 2GB limit of the Samba file system. With the current setting, 2GB is equivalent to around 45 minutes of video. A good benefit of this file split is that we can start watching a show as soon as the back-end starts recording the 2nd file.

More information available on the WVR project page.

This is a bug fixes only release.

More information available on the WVR project page.

Added the ‘Files’ menu to access and delete video files. Fixed several bugs/enhancements.

More information available on the WVR project page.

First release with the check for overlapping programs. The front-end will not let you program two recordings that are conflicting anymore. Several bugs fixes and enhancements were included in this second release.

More information available on the WVR project page.

First public release of WVR, a web front-end for scheduling the recording of one-time and recuring TV programming.

This is the first working version. Some major functionalities are still missing.

More information available on the WVR project page.

I have been using Coyote Linux as my router for a few years now. This is a very good way to reuse an old (486 based) diskless PC into a nice router which can be configured and customized as much as you can code patches and scripts (this is a minimalist Linux system) and make them fit on the boot/root floppy.

One issue with Coyote Linux 1.31 is that the IP Masquerading module for supporting H323 voice and video calls is no longer handling the incoming calls. Only calls initiated from the LAN side are properly supported.

I’ve put together a replacement module here as a drop in replacement of the delivered module with 1.31. A README included in the archive provides information on how to install and configure the module.

www.ljmsite.com/projects/coyotelinux/