I’ve been tackling this question lately. I’ve been a generalist for many years, giving me a chance to dive into technologies as varied as PHP (early career), Java, C#/ASP.NET, Ruby/Rails, iOS, and Javascript (Ember.js).

This ability has served me really well in past projects where I’ve worked with a number of smaller companies where you have multiple responsibilities requiring proficiency across multiple technologies.

There are upsides and downsides to generalization. The upside is you are always learning new ideas if you keep an eye out. As with natural languages, knowing many technologies helps you learn new technologies even faster. This can be especially helpful when your toolbox of skills doesn’t have the right tool for a particular problem. The downsides are that you can reach a limit of how deeply you delve into an ecosystem if you are in and out of that world.

There are many benefits in specialization. You get a chance to deeply understand a subject and forge new ground in that discipline, which can be truly rewarding. But there are risks as well. If you pick a technology that eventually fizzles out, you could be stuck. You also run the risk of running into a smaller variety of problems, bringing down your possible creativity when solving problems that aren’t a good fit for the technology you specialize in (like doing systems programming in Ruby).

But after some introspection lately, I’ve come to a surprising realization — to generalize or specialize is not mutually exclusive. I think a good balance for someone like me could be to specialize in a set of concepts or class of problems, yet generalize with technology implementation of solutions for those problems.

One example is web performance. Optimizing for front-end web performance can involve optimizing downloads, concatenating scripts, and also thinking what to load in optimizing the user experience. But whether you’re applying those concepts to Apache or nginx, or Ruby on Rails or ASP.NET MVC, you are still accomplishing the same goal, but with different technologies. The real value is in knowing what solutions to apply. Once you know what you’re looking for, it’s easier to find out how to do that particular task in a particular technology. Specializing in this problem set requires knowledge of upcoming solutions like the adoption roadmap for HTTP/2, and optimizations to reduce SSL overhead.

Another example is architecting for High Availability. Yes you can get highly specialized in this industry depending on the stack (knowing the difference between MongoDB Replica Sets vs SQL Server AlwaysOn), but knowing the pros and cons of each can help people make decisions. You can also visit https://www.couchbase.com/products/mobile to get more details.

Of course, there are some technologies that pair well with some industries. A lot of big data work uses the JVM (Hadoop, Spark), a lot of web frameworks tend to shy towards dynamic languages like Ruby and Python. Systems programming has tended to follow a C/C++ route, although lately is seeing an influx of dynamic language developers thanks to languages like Go and Rust.

But I think there can be great value in specializing in how to solve a class of problems, but knowing how to solve them in a handful of different, yet common stacks.

When my cousin, a liberal arts major, got a Mac based on my recommendation (the first of many experiments where I’d use my loved ones as guinea pigs) and tech support questions all but disappeared, I started paying attention.

When my brother, a History major who loves jokes about Henry Kissinger, bought an iPod and had been using it regularly without ever asking me what he should buy or how he should set it up, I was impressed.

When my wife, a conservation journalist, made a professionally looking hard-bound photo album of our wedding, I was delighted.

When my uncle setup email on an iPhone without any guidance from me, the first time he ever used it, I was thrilled.

And finally, when my aunt easily learned how to text on a hand-me-down iPhone after years of staying away from cell phones, I was no longer surprised.

None of these people have any interest in being technologists. They don’t talk about clock speeds, Anti-Virus, or defragging. They talk about reading, planting virtual crops, and enjoying movies, best EMF protection products and spider solitaire in bed.

All my life, I never understood what it was like to be a fan of anything like a band or a sports team. But watching the people I love enrich their lives in small ways with technology without needing my help, I became a fan.

Thank you Steve for inspiring us, by example, to build technology that disappears as it delights.

802.11b, the original standard with mainstream success, was good enough for web browsing but not much more. Its successor, 802.11g, has been good enough for web browsing and streaming most kinds of video, and is still the most commonly deployed standard in today’s home routers and mobile devices (including smartphones). Your home is a place to showcase your personality and style, you can get some advises of how get a nice design in home art, https://www.urdesignmag.com/architecture/2020/08/19/ways-to-incorporate-art-into-your-home-decor/.

But newer 802.11n standard routers are available this year (and supported on most new laptop models) that allow you to reliably stream blu-ray quality HD video anywhere in your home (with a strong enough signal permitting). But to get the sort of reliable throughput you expect for uninterrupted HD video streaming, a few tweaks need to be made.

Enter Apple’s Airport Extreme.

Apple’s gear is generally quite good, made from quality parts, and speedy out of the box. But as a consequence of how 802.11 networking works, you can make a few tweaks to the Airport Extreme configuration to get extra speeds.

802.11b and g traditionally ran on 2.4GHz networks. But 802.11n added 5GHz to its operating frequency. Unlike the 2.4GHz band, 5GHz offers much less interference, at least for now, only because there are much fewer devices that run on them. But this ability also gives wifi routes additional power to run at much higher speeds.

However, for maximizing backwards compatibility, if there are any 802.11b/g devices on the N network, the router kicks back into G mode, and runs on G speeds. To get the most out of your N network, you need to run a separate network purely for N devices.

Since you are running a network on N devices, and since all N devices support 5GHz, why not run it on 5GHz? This is exactly what the Airport Extreme gives you, but in a very non-obvious way.

Note below the transfer speeds I measured from transferring a 1GB video file across the G, N (2.4ghz) and N (5GHz) network:

File Size: 1.01 GB h.264 video

Sharing using the AFP protocol between a Mac Mini (source) and Macbook Pro (destination) 8 feet away in the same room:

802.11g at 2.4 GHz: 769 seconds

802.11n at 2.4 GHz: 467 seconds

802.11n at 5 GHz: 209 seconds

The huge difference in running on a pure 5 GHz network is worth the upgrade to a good 802.11n router. The Airport Extreme allows you to run two networks (one hybrid, one pure 5 GHz). Definitely exercise this option so your N devices can run on their own network and get the huge (100%+ speedup) benefits.

After examining the UI, I had some concerns that’ll hopefully be addressed by Google.

Firstly, it is a very slick UI. Everything moves smoothly like butter,  and it is a very nifty UI for demonstrations and showcases. But is it an effective way to browse news articles?

When I browse for articles with heavy textual content (like in most newspapers or even Google News), I look for headlines. In the default view of fastflip where the thumbnails are very small, the headlines are too small to read. You could read them if you squinted a bit, but that would make things harder, not easier, to browse for articles. To address this (and probably accessibility), Google displays the headline of the article under the thumbnail. However, if my eyes have to keep shuttling back and forth between the thumbnail and headline, that results in a jarring effect. Yet more work I have to do on an interface that’s supposed to be more “casual.”

Visual Artwork based browsing

The idea of replicating the flipping of magazine pages is interesting. But most pages in magazines are meant to catch your eye. Whether they use beautiful artwork in advertisements, or provocative messaging, the idea is that you have about 1-2 seconds of a reader’s attention, so you better catch their attention enough to want to find out what you’re selling to them.

But this idea doesn’t work on the web, or with fastflip. Most of the articles are bodies of black text on white backgrounds. There is simply not much distinguishing each article, apart from the design of the publication’s website, and occasionally a really beautiful picture which will draw your attention.

On the topic of design and artwork in newsstands, there’s a great TED Talk.

Perhaps some of these designs would make Google’s FastFlip more compelling.

java.net.BindException: Address already in use: JVM_Bind 
Cause: java.rmi.server.ExportException: Port already in use: 1098;

The first (and more obvious) solution is to check that there isn’t already an instance of JBoss. Once you’ve made sure (by killing all java.exe/javaw.exe processes), how do you find out what program is still listening on a particular port?

Netstat will help you there.

Netstat

Netstat (or equivalent) is a utility available in all major OSs, and allows you to view protocol statistics.  You can view incoming and outgoing ports that are in use, along with the IPs and hostnames that the sockets are connected to. In Windows, typing netstat -ab (-a for all connections, -b for the executable involved) in the command prompt results in a list of TCP and UDP sockets and the name of the process that is listening on that socket. Here’s some example output:

  TCP    MyComputerName:6060    MyComputerName:0       LISTENING       1132
  [ibmtcsd.exe]

  TCP    MyComputerName:1026    MyComputerName:0       LISTENING       1508
  [UNS.exe]

  TCP    MyComputerName:1141    MyComputerName:0       LISTENING       28668
  [googletalkplugin.exe]

  TCP    MyComputerName:5152    MyComputerName:0       LISTENING       1572
  [jqs.exe]

  TCP    MyComputerName:5354    MyComputerName:0       LISTENING       896
  [mDNSResponder.exe]

  TCP    MyComputerName:27015   MyComputerName:0       LISTENING       12500
  [AppleMobileDeviceService.exe]

  TCP    MyComputerName:netbios-ssn  MyComputerName:0       LISTENING       4
  [System]

  TCP    MyComputerName:1141    localhost:1142         ESTABLISHED     28668
  [googletalkplugin.exe]

  TCP    MyComputerName:1142    localhost:1141         ESTABLISHED     29716
  [chrome.exe]

  TCP    MyComputerName:1405    localhost:1406         ESTABLISHED     33508
  [firefox.exe]

Hope this helps you track down that annoying process that’s keeping JBoss from loading.

• Edit files on the server in a secure way
• Use Textmate or a local editor
• Not have to run rsync or any sync/upload utility

Enter MacFUSE (and MacFusion) and SSHFS.

MacFUSE is an OS X port of the excellent FUSE (Filesystem in User Space) project. It allows you to build your own “filesystem” quite easily. (technical video on MacFuse available) Some nifty filesystems include:

• SSHFS – allows you to mount a directory on a remote server over SSH, as a local Volume
• FTPFS – same, but using FTP

The beauty of this is, programs that are only designed to edit local files can now edit remote files without realizing it.

Step 1: Mount the remote folder as a local volume

First, download MacFusion, which allows you to setup your SSH server parameters and mount it as a local volume. I was initially going to type out sshfs shell commands, but this is a much cleaner solution.

MacFusion should be intuitive enough that you can enter the host, username, password, and main directory. For extra security (yes, this actually is more secure), use public key authentication for SSH with the added convenience of auto-login (if your certificate doesn’t have a password).

Step 2: Edit in Textmate (or other editor)

Textmate (TM) is slow while using SSHFS for various reasons.

Firstly, when TM loses focus (say, you activate your browser) and regains focus, it rescans your project directory for any file changes. While this is a nifty and near-instantaneous feature for a local volume, it will take a few seconds on a remote server. There’s a quick remedy for that: ReMate – a TM plugin that disables the project folder refresh. This will make the experience much more bearable.

Saving files will result in a noticeable lag, but I think this is overall a good solution for having your cake files on a remote server) and eating it too.

Anyone inclined to suggest source control or rails like capistrano deployment scripts, this is a cbd products and dirty solution for wordpress themes and editing files on the server for instant feedback. For more complex apps, those deployment packages will do a much better job.

• Visited the farmers’ markets in the Eastern Market/Capitol Hill area
• Cherry Blossom Festival at the Tidal Basin
• Jazz on U Street (Bohemian Caverns, unique atmosphere, good food, great music)

It made me think about all the other things I haven’t discovered in this amazing city and the surrounding area, which inspired me to compile a sort of “bucket list” but for the summer. This is going to be a live list that I’d like to tick off and blog about as I accomplish these things.

The criteria for the list are as follows:

• Must happen in the DC Metro Area (or within a 2 hour drive)
• Must not take more than a day to complete (most activities will be a few hours at most)
• Indoor or Outdoor, doesn’t matter
• Preferably something more unique to DC (watching movies doesn’t count)
• If I haven’t done it in 2-3 years, it counts.

Without further ado, the list (in no particular order):

• Ride the Capital Crescent Bike Trail
• Ride the Beach Drive Bike Trail
• Ride the Mount Vernon Bike Trail
• At least 1 restaurant on DCFoodies’ list of favorites
• Go to the top of the Washington Monument
• White House (sober, during the day, and not when on one of my introspective drives at 2am)
• National Air and Space Museum’s Udvar-Hazy Center (near Dulles Airport)
• Dupont Circle Farmers Market
• Outdoor concerts by the Army, Navy, Air Force and Marine Corps Bands at the Capitol grounds
• (From Limbo in comments) USMC Band at the Barracks on I St, and at the Iwo Jima Memorial (Tuesdays)
• Performance at the Carter-Barron Amphitheater on 16th St
• Performance at the Kennedy Center (I used to go 4-5 times a year, haven’t gone in ages)
• Performance at the 9:30 club (seeing Prodigy there on May 18)
• Performance at Wolftrap
• Grab a Telescope and go see the stars at night in Shenandoah Valley
• Ethiopian Food on U St
• At least one more Jazz club on U St (apart from my regular Utopia and now, Bohemian Caverns)

Suggestions on things to add?

1. My own company’s system hosted on Freshbooks – This is the master copy since my invoicing is driven by it
2. Quickbooks, two of my clients use it and require me to enter hours in there for accurate billing
3. GHG’s Timesheet software – err… this abomination of an app barely functions, forget asking for an API

I want to be able to record hours on all 3 systems in one click or even from my iPhone.

Freshbooks

Freshbooks’ API is very clean and they have an iPhone app, so this is the least of my worries.

Request:

<request method="time_entry.create">
<time_entry>
<project_id>1</project_id>        # Associated project_id (Required)
<task_id>1</task_id>              # Associated task_id    (Required)
<hours>4.5</hours>                # (Optional)
<notes>Freshbooks Test</notes> # (Optional)
<date>2007-01-01</date>           # (Optional)
</time_entry>
</request>

Response:

<response status="ok">
<time_entry_id>211</time_entry_id>
</response>

Almost too easy. Thank you Freshbooks.

Quickbooks (the Big Kahuna)

Quickbooks doesn’t have an API, but have exposed API-like functionality to their Outlook plugin and Vista Gadget. Since Vista Gadgets (and Apple Dashboard Widgets as well) are just html/javascript, some simple hacking revealed exactly what I needed.

“Extracting” the Quickbooks Time Tracker Vista Gadget’s protocol

I first downloaded the .gadget file from Quickbooks’ site after logging in. Since this is a CAB (Cabinet) file, you can open it natively within Windows or use File Juicer on the Mac. The CAB file reveals a directory structure with tons of images, html files, and javascript files. From here, it’s relatively easy to find the right files you need.

Of particular interest are:

js/loginAndSynch.js
js/persistentSettings.js

After about half an hour of hacking at the javascript code and many invalid cURL requests to the QB server, this is essentially the protocol I was able to get working:

“GetLists” – Obtains the list of customers and service jobs from the Quickbooks server

Endpoint: https://timetracking.quickbooks.com/j/tts/widget

POST Body:

cmd=getLists&version=1.0v&login=%7B%22companyid%22%3A%22your_company_id%22%2C%22username%22%3A%22your_username_value%22%2C%22password%22%3A%22your_password_value%22%7D

cURL command for debugging: curl -v -d @post.data -X POST https://timetracking.quickbooks.com/j/tts/widget

Make sure you put the POST body above in a file named “post.data” in the same directory where you run cURL. It’ll use the contents of the file as the POST body.

And you will receive all the Customer Names and Service Items in JSON (fake company names/IDs for confidentiality):

{"CustomerJobs":[{"Name":"Food Network","ID":"999999"},{"Name":"Food Network:Rachael Ray","ID":"7777777"},{"Name":"Cartoon Network","ID":"25252525"},{"Name":"Cartoon Network:Family Guy","ID":"2343333"},{"Name":"Vacation","ID":"888888"}],"ServiceItems":[{"Name":"Contractor-Admin work","ID":"111111"},{"Name":"Contractor-Support Work","ID":"222222"},{"Name":"Contractor-Support Work:Training","ID":"33333"},{"Name":"Support Work","ID":"444444"},{"Name":"Support Work:Maintenance Work","ID":"5555555"}]}

The easiest way for me to quickly inspect a JSON struct is to fire up the Ruby interactive console (irb) and do the following:

>> require 'rubygems'
=> false
>> require 'json' # 'sudo gem install json' if you don't have the json gem
=> true
>> a = JSON.parse(File.read('customers.json'))

… ‘a’ now has the json struct

From here, I can iterate through the hash and inspect its keys and data:

>> a.keys
=> ["ServiceItems", "CustomerJobs"]
>> a['CustomerJobs'].each do |job| puts job['Name'] end

… prints out all the customer job names

Voila! We have the first half (and arguably the easier half) of this Quickbooks API project.

In the next part, I will outline how I did the second half, the actual Time Entry piece. (After I figure it out)

illegal character

Illegal Character error in Firebug

The last time I encountered this problem, I renamed the file a few times and that fixed it. Yes, that makes no sense but as long as it was fixed, I just went about my merry self until it happened again today, and no amount of file renaming would fix it. So I did a little bit of investigation to figure out exactly why this happened.

First, I found this problem only happening in Firefox, and not Chrome, or IE.

When accessing the file directly in the browser, the first 3 chars were:


/*
*
*
*
*
*/

(function($) {


cURL returned the same result, narrowing it down to either the server adding that character, or the file containing it. Since the server is JBoss, it’s most likely the file.

Opening the file in Binary mode in Textpad, I found that the first 3 bytes had the hex value: EF BB BF

This has to be a standard hex header for a certain file format (probably unicode related?) So I googled for that hex value. The first result was a wikipedia entry for Byte-order mark. This made total sense already. The Byte-order mark specifies the byte order (endianness) of the file that follows, with relation to unicode-related encodings. It turns out that EF BB BF is the Byte-order mark to denote UTF-8, but at the same time, it is completely unnecessary as UTF-8 has no byte-order issues by design.

So I had to get notepad++ to remove the first 3 bytes of that file, by choosing the option for “Encode in UTF-8 without BOM” in the Format Menu.

This fixed the issue! I hope this helps others that encounter this completely obscure looking error.

So for example, if you did the following in the Python console:


>>> a = ['a']
>>> dir(a)
['__add__', '__class__', '__contains__', '__delattr__', '__delitem__', '__delslice__', '__doc__', '__eq__', '__ge__', '__getattribute__', '__getitem__', '__getslice__', '__gt__', '__hash__', '__iadd__', '__imul__', '__init__', '__iter__', '__le__', '__len__', '__lt__', '__mul__', '__ne__', '__new__', '__reduce__', '__reduce_ex__', '__repr__', '__reversed__', '__rmul__', '__setattr__', '__setitem__', '__setslice__', '__str__', 'append', 'count', 'extend', 'index', 'insert', 'pop', 'remove', 'reverse', 'sort']


As you can see, the object ‘a’ is an array. And when I type dir(a), the python console returns the properties available to a Python Array object. It’s an extremely useful way of discovering the features of the languages, especially when you just like tinkering.

It turns out (and I didn’t know this) that there’s a similar command in the Ruby interative console, help(). This basically allows you to look up the documentation for base classes and features of the given class. So you can do the following (slightly different from Python):


>> help(Array)


Ruby sends you to a text page similar to a *nix man (manual) page. This works on the class level and also the class method level, such as:


>> help('Array.delete')


Note that when mentioning a method for a class, you must enclose it in quotes, or it won’t be able to find it in help. It’s not necessary to enclose a class lookup in quotes though. This is a really useful command for people who are stuck on a plane or in a similar offline situation, like I am right now. It can also be much quicker than googling for a particular reference, as you can keep one terminal shell open with irb running all the time, and get your reference in an instant.

See what you can discover just by tinkering with help() or dir() commands in Ruby and Python.