I just got a 2nd access point working at home.  It’s about 50ft away from the existing access point.  I had this working before, but replaced the main router (replaced dLink with Belkin) and it stopped working. 

I was able to rule the problem down to a cable running about 20ft from my main router to the 2nd router. 

• If I plugged the 2nd router into a 4ft patch cable it worked fine.
• If I plugged the 2nd router into the 20ft cable it didn’t work at all – no lights on either router, no ping response, nothing
• If I put a continuity tester on the 20ft cable it showed all 8 wires were connected in the right order.

So – Short cable worked fine – long cable didn’t work at all.

After much research and a call to Belkin (who said 15ft is the max cable length) I solved the problem.  It boils down to what I consider a dirty little secret of network cabling:

When creating a cat5 cable you must wire the ends such that twisted pairs end on specific pins.  It is not enough to just get the two ends wired the same. Pins 1&2 must be a twisted pair and pins 3&6 must be a twisted pair.

Here’s the dirty part:

Those are the only 4 pins you need to hook up.

What it means is you wire both ends something like this

Pin1: Orange Wire

Pin2: Orange & White striped Wire

Pin3: Green Wire

Pin6: Green & White striped Wire

And you leave the rest empty.

More info on 1000bast T cables (which need all at this link.

I know I have asked network guys and cable pulling guys specifically if there were any tricks to wiring the plugs, because I suspected something like this for a long time – and they all refused to tell me.  I vaguely remember one or two of them having an evil grin when they told me this.  Job security?  Maybe.  It was hard to learn, it should be hard for you to learn kinda thing?  More likely.

We moved a year ago and in that time I think I’ve had to replace about 4 light bulbs in the new house.  In the old house I was replacing what seemed like 2 or 3 a month.  I’m not sure if it’s a property of the wiring, or some quality of service from the electric company. 

I don’t think I’ve had to replace a single CFL bulb in either house.

Bought a new router (net connection was dropping for a second a few times a day) and replaced it with a Belkin N Wireless Router.  All hard wired computers were back online pretty quick.

The Airport Express that I use to play music in the office did NOT connect though.  When I tried to configure it I would get this:

An error occured while trying to access the Apple wireless device.  Make sure your network connection is valid and try again.  Error 10057

After trying MANY things I flipped my router from “802.11b & 802.11g & 802.11n” to “802.11g” the Airport Express now works.  I don’t know what this does to my range/speed for devices that are able to talk 802.11n.

The bigger problem is the long network cable that no longer works… zero changes to that end of things but connecting a computer directly to the other end fails to work.  Little choice but replacing the entire cable, about 20ft long… through the basement… through a hub.

I hate networking – but not as much as I hate printers.

The basic idea: A method of presenting information such that you get as much detail as you need based on how long you read/view the info.

Example one: The Red Sox beat the Orioles last night… the key piece of info is who won.  So one of the smallest versions of this story could be “Red Sox Win.”  Most people will want more – like the score, maybe they didn’t know who they were playing or where… so another version is the obvious “Boston Beats Baltimore 3-1 at Fenway.”

For purely academic discussion: If the news query was “How did the Red Sox do last night?”  The answer could be boiled down to a single bit.  1= They Won, 0 = They Lost. If all I had time for was a single bit – this would be a great piece of information.

Continuing the academic discussion: What if I had time for 2 bits?  Maybe there’s nothing useful in a 2nd bit… or maybe it could be defined as “At home?”  So last nights story in two bits is 11, which translates to “They Won, and they played at home.”

The score and opponent are usually viewed as important pieces, so we have to define how many bits for this info, and we eventually get to a “11111101″ = which breaks down to “They won, at home, 3 to 1 against the 4th team in a list of MLB eastern league teams.  An algorithm could be designed to fit useful info into as few bits as possible.

Back to the big idea – currently we have a very established system of “Head Line” and “Story.”  I’m sure there’s a commonly accepted maximum length for the Head Line – maybe 100 characters, and the Story tends to range from a front page “couple paragraphs” to magazine styled “as much as can be read in a typical bathroom session.”

My idea is to enable an author to write a story such that it can be consumed in almost any size.  The Sports story is easy to define – but where this really hits the fan is international political stories where the audience has a questionable amount of previous knowledge.  Someone very up on their geopolitical events could get a short version, but someone without this background could consume more bits to deepen their understanding.

Weather is another topic that could probably be broken down to the single bit response to “Will it be nice out tomorrow?” to the more complex, “What temperature will it be during school hours?”

Our house has at least 10 sliding doors.  I think 2 of them work like they are supposed to.  Last night the sliding door on the shower decided to go off track and I can’t get it back on.  Found out the sliding door in the bedroom is letting mosquitos in.  I’m looking for something to replace all the sliding screen doors, as almost every one I’ve owned in my life has been problematic.

I installed my map in my office back in September, and just read a blog post detailing NatGeo giving the same map to the President as a gift.  Although his is actually a map cabinet that holds lots of maps.  Kinda kewl to have pix of my office and the oval office in the same blog.

After upgrading to v3.0 of the iPhone operating system the eReader Pro application can no longer switch what book your reading.  A big problem.  The reader still worked, but if you finished a book (as I did last night) and tried to open another one you’re kinda just lost… it didn’t crash, it just didn’t go to the bookshelf.

To the developer’s credit their company, Fictionwise responded to an email I sent complaining with this:

We apologize for the trouble. A fix is currently in progress. We do not have a definite time for when it will be made available at this time. But it should be soon.

The most surprising thing about the response is that I got it about 5min after sending the complaint. Expected response from this kind of thing is days, not minutes.

I found reports of this problem in a couple places, but only buried in the iTune store application reviews did I find the work around: Delete eReader Pro and re-Install it.  This gives you a chance to open a new book – but has the side effect of deleting all the books you’ve downloaded as well as resetting your application preferences.

Had I got this from fictionwise I would have been much happier.  Ideally I would have liked to find a great big, top of the page kinda support link on their help pages detailing the problem and the work around.

Even though there’s no official link to anything from China, I like the Chinese Curse – it gives you a lot to think about:

• May you live in interesting times
• May you come to the attention of those in authority
• May you find what you are looking for

I’ve been encouraging the boy to understand how things work – and decided a fun way to do that would be to take things apart, after all that’s what I did at his age.  So we headed to the Flea at MIT on a hunch they might have some fun stuff to take apart.

Turns out it was the perfect place to find exactly what I had in mind.  The place is a geeks dream come true.  So much old technology I didn’t know where to start.  I was tempted to buy an old oscilloscope, as their were at least a hundred of them in different stalls. 

I ended up finding a reel-to-reel tape recorder a guy was selling for $50.  I started talking him down on price and asked, “Does it still work?”

He replied, “Perfectly.”

I told him the plan was for me and my son to take it apart, and that I didn’t want to ruin something that still worked well.  He agreed and told me to take the reel-to-reel recorder sitting next to it for free.  It weighed 30pounds easy and I had to carry it all over the place.

Nick really enjoyed it – although maybe next time I’ll get something that we can see working first and maybe keep plugged in (low volt DC something) to watch it work.

Anna was along for the whole ride – and I had a screw driver ready for her.  Instead she played repoter with the stuffed weasel beanie baby she got at the Flea at MIT.

They really don’t make things like this any more – hardly any plastic.

The kewlest part of this device in my mind was that it had vacuum tubes.  I was tempted to try to get the amplifier part of the thing working and maybe build a small guitar amp or something… but my ambition escaped me.

Here’s a pretty kewl shot with the tubes visible:

We got it apart as much as we good – with a ton of screws and bolts lying around.  The carcase sat on the table for a week or so – hard to dispose of that much metal.

All in all the project was a complete success.  Quality dad-kids time and maybe a little education.

I wanted a method to show what the version or subversion revision number a user was currently running.  The main reason is to determine what code base was being run by users reporting problems I thought we fixed, or not seeing functionality I thought was complete.

Most people are familiar with this functionality exposed via the “Help->About” menu on a desktop application.  Web developers need to do this differently, as there typically is no outer application menu. 

Being a flex application, I have the ability to modify the context menu users see when they right click.  This was part one of the puzzle, and that answer is well documented in this post by Kenny Lin.  The way I have my SDK wired, much of the code documented there did not trigger the autocomplete on my system, but it compiles and runs perfectly.

Getting the revision number is the hard part.  We currently use subversion as our source code repository, and it has a global revision number – what most developers would refer to as the build number. (Due to the fact that we now do a build every time we do a subversion update, see below.)

Subversion has a technique called keyword substitution which allows you to put a string in a file like $Revision$ and every time that file is updated on the repository subversion will update that string with the correct number.  I thought I was done right here – but the problem is that subversion only does this substitution when it updates a file, i.e. when a developer edits it.  I added this to our main application file, and it worked the first time I updated it, but it turns out we don’t update that file very often anymore.

The biggest piece of the puzzle was an autobuild process which is triggered by subversion updates.  My very simple build script generates a subversion info file using this shell script command:

svn info > assets/svnInfo.txt

I embed this file in the application via this technique which is only a couple lines of actionscript code:

//Embed a text file that is autogenerated by the build process
//it contains the revsion number of the project from subversion
[Embed("assets/svnInfo.txt", mimeType="application/octet-stream")]
private static const svnInfoText: Class;

And the following function parses the revision and puts it in the context menu:

/**
* add the version number from the subversion generated file
* to the context menu so testers can verify what
* build they are running
*
**/
private function setUpContextMenu():void
{
    var myMenu:ContextMenu = new ContextMenu();
    myMenu.hideBuiltInItems();
    var svnTextInstance: Object = new svnInfoText();
    var svnString:String = svnTextInstance.toString();
    var findString:String = “Last Changed Rev:”;
    var revisionString:String = svnString.substr(svnString.indexOf(findString) + findString.length +1, 5);
    var menuItem1:ContextMenuItem = new ContextMenuItem(“MyApp v0.1.”+ revisionString);
    myMenu.customItems.push(menuItem1);
    this.contextMenu = myMenu;
}

Getting the autobuild process up and running was the hardest part – learning the various libraries I needed and setting up the license information from adobe.  I have a lot of ideas and opportunities around that, including ANT or other build tools – but for now the simple fact that I can control the build to the level of adding this step of creating the subversion information file get’s me the functionality required.