" /> Mindless Fluffiness: November 2004 Archives

« October 2004 | Main | December 2004 »

November 30, 2004

Mobile phones kill early media? One can only hope

I love the mobile phones guys - first they trained users that lousy voice quality really was OK. Then they got people to understand that when they were done pressing all the digits in a phone number, they had to press some some other send key before anything would happen.


Now they have an approach to eliminate any one way media from the far end during the "ringback" time, and of course an desperate attempt to try and wrap a business model around it. See http://apnews.myway.com//article/20041129/D86LODF00.html for a story on charging for songs that play as your custom "rickback tone".

This of course would be a huge step forward for the phone industry. The current approach of you hear the "busy" tone from the country you are calling is insane. How do I know what "busy" sounds like when I phone Antarctica. This whole practice is the exact opposite of what Internationalization of most technology tries to achieve. When I make a call, I should get a consistent indication of "busy" that I understand. The funny thing is that it is incredibly difficult to achieve this useless feature and the only argument for it is basically "that's how it has always been done on the PSTN" and "users would not accept any change". People regularly tell me they actually know tones - this seems unlikely, there are quite a few. Check out this list David and I did.

It's not quite true it this early media has no use. It is also used for basically toll fraud purposes. Many large IVR system (1-800-gofedex being the canonical example but including many other people like the US IRS) use it to play out the initial prompts of their IVR as "ringback" so they can delay connecting for a little extra time.


Update from Adam Roach

"My reading of this is that it affects the ringback that people hear when
they call *you*, not the ringback you hear when you call people. In other words, this virtually guarantees early media for a long time, since it's now a money-making service."

Did I mention the mobile guys really drive me nuts.

--------

November 24, 2004

Photos from ReSiprocate Nov 04 coding session

These are the photos of the white board - it will only make sense to people who where there. Actually, I doubt it will make sense to them either but people asked me to post them.

--------

Another happy OpenSSL users

Rohan leans the difference between add_all_algorithm and add_ssl_algorithms
9:15 AM - Rohan starts hacking up a program that uses OpenSSL



10:15 and starts reading the documentation, and comes to the conclusion it sucks



11:15 AM - None the less he happily finishes his program. This openSSL stuff is not so bad after all, but the documentation still sucks.



then decides to run it.

6 pm - still does not run and all functions return success



9 pm - still no joy



11 pm - Rohan has become demonic and is bending the space time around his ass.



11:00 pm - A real expert arrives on the scene



11:02 pm - s/ssl/all and the program works.

Seems that more things that just the documentation suck.

--------

November 19, 2004

Free as in beer? Not a Cisco beer!

At IETF 61 I had the opportunity to buy the most expensive beer I have ever purchased. It was a Cisco beer.

--------

November 12, 2004

Computer Vision for Line Drawings

May 1993 MSC Thesis at the University of Calgary by Cullen Jennings
Modern maps and engineering diagrams are usually constructed and stored using GIS or CAD systems. A large number of drawings, however, exists only in a paper form. This thesis examines the problem of automatically converting such drawings and maps from raster image to high quality vector GIS or CAD forms.

This thesis begins with a review of previous work in the area and then proposes a new method based on findings about how human vision works and domain specific knowledge. Another system based on the classical work in this area is presented, to which the new system is compared. This comparison shows that the method proposed here obtains substantially better results than classical methods. The time a human operator could expect to spend correcting the errors created by this system would be less than one tenth of the time required to correct the errors created by a classical vectorization system.

http://www.dial911anddie.com/papers/msc_thesis.pdf

--------

November 06, 2004

Clean up photographs of text documents in PhotoShop


Download CleanDocs.atn from http://www.libvoip.com/e911/cleanDocs/CleanDocs.atn

Start Photoshop

Go to the Window menu in photoshop and make sure that the action palette is visible.

On the action palette, on the top right there is a circle with an arrow on it. Click on this menu and select "Load Actions"

Find the CleanDocs.atn file you downloaded and load it

Now you can load a photo, click on the CleanDocument action, and hit the play like arrow on the bottom of the action menu to see what it does.

Clean and Save cleans it and saves it - unfortunately it tires to save it in a file C:\Documents and Settings\Cullen\Desktop. You can change this be editing the action. Click on the arrow beside the action to expand it and you can see the save command part-way down. Double click on this (when you have an image loaded) to save it somewhere different and change the action.

Go to the File menu, then Automation, then create Droplet and create a droplet that has the Set as "CleanDocs" and the action as "CleanAndSavePrint" and uncheck all three check boxes under it. This will create a droplet on your desktop that when you drag an image over on top of it and drop it, it will Clean it, Save it, and Print it. You can do similar thing to create a droplet with the action set at CleanAndSave.

You can use the File, Automate, Batch processing stuff with this action too.

--------