Something I'd like to see:

A site that maintains all the world's 'lookup tables.'

Every lookup table you'll ever need, exposed in open, compatible and readily-consumable ways.

All lists can be retrieved as html or xml, as json, as csv (the one format guaranteed to outlive the cockroach), or any other commonly used** data format.

Lists at 'wiki lists' are more complete, more accessible, and much lighter-weight than the Lists at wikipedia.

The lists are crowd-sourced, contributed by anyone and curated by the detail-oriented masses. The faceless trainspotters and plane spotters, the aspies and cybrarians.

Lists that are currently shown in haphazard ways inside wikipedia could instead source their content from such wiki-lists.

Where a list can be reliably sourced from a canonical source, it is done so programmatically and the attribution provided in a standard way.

Example lists include...

• Every sovereign nation
• Their dialing code, currencies, major languages
• The three letter codes for every airport, landing strip and helipad in the world
• The name and abbreviation of every state in the united states (and every other nation)
• Zip codes and postal codes
• Monarchs of England
• Breeds of dog
• Keywords in C
• Mayors of southhampton
• Notes in F major pentatonic
• Winners of the Belgian grand prix
• Stocks that trade on the NYSE

And a thousands other reference lists. Short lists. Information that changes slowly, slow enough for humans to curate it.

And they're the kind of lists that you (or someone like you) needs to reference in their applications and research all the time.

(See also this thread at Hacker News which prompted me to write down this idea I've thought for so long.)

** Note I said 'commonly used' rather than 'standard'. CSV is commonly used, but it doesn't have a standard. Oh you might point out that there are standards for CSV, and of course there are. Where there's standards, plural, there is no 'standard'.

When you've done everything you can with Ruby On Rails....

When you've reached the limits of Haskell on a Horse...

If FubuMVC is not for you...

It's time to meet the MVC web framework to end all MVC Web frameworks.

DOS on DOPE is the modern MVC framework built on the awesome power of Batch scripts.

All of the controllers in DoD are batch files.

All of the views are batch files.

The model is based on batch files. The helper functions are... you guessed it! Beautiful batch files!

The complete source code and tutorials are available at github

With DosOnDope, it's this simple:

open a cmd window, and write:

dope site Blog

You just built a website, called 'blog'. Here it is:

Oops. It moves too fast, I can't keep up.

Here's all the folders that have been created for us...

C:\temp\blog>dir /s/a:d/b
C:\temp\blog\Content
C:\temp\blog\Controllers
C:\temp\blog\h
C:\temp\blog\Models
C:\temp\blog\Scripts
C:\temp\blog\Views
C:\temp\blog\Controllers\Home
C:\temp\blog\Views\Shared

And all the files...

C:\temp\blog>dir /s/a:-d/b
C:\temp\blog\Default.aspx
C:\temp\blog\favicon.ico
C:\temp\blog\Global.asax
C:\temp\blog\web.config
C:\temp\blog\Content\Site.css
C:\temp\blog\Content\li.png
C:\temp\blog\Content\Logo.png
C:\temp\blog\Controllers\_actionMissing.bat
C:\temp\blog\Controllers\_controllerMissing.bat
C:\temp\blog\Controllers\Home\Index.bat
C:\temp\blog\h\a.bat
C:\temp\blog\h\h1.bat
C:\temp\blog\h\h2.bat
C:\temp\blog\h\h3.bat
C:\temp\blog\h\p.bat
C:\temp\blog\Models\split.vbs
C:\temp\blog\Views\Shared\_header.bat

Let's create a new controller action, and see how it looks.

Let's improve the look of that page. You type a little...

...hit refresh and see what changed.

It worked.

Notice the use of html helper methods like 'p.bat' and use of the partial, '_header.bat' to give styling and a title to the page.

But websites don't live in a vacuum. We need some data to bring this thing to life.

In DoD we embrace the NoSQL movement and jump straight to the data-store of the future: a CSV file.

C:\temp>CD Blog\Models
C:\temp\Blog\Models>copy con Posts.csv
ID`Title`Article
^Z

Posts.csv is both the model that describes our data, and the store which will contain the data itself.

Let's generate actions for administering our Posts:

C:\temp\blog\Models>dope gen Post
Generating actions for MODEL Post in Controller Home...

Check in a browser...

The generator has created a nice page for us, with a name and textbox for every field, so we can create new posts.

Once we've added a few articles, let's see what the index looks like:

If we look inside the posts.csv file, we see the data:

C:\temp\blog\Models>type Posts.csv
ID`Title`Article
1`Hello Brazil!`Yes, yes, hello indeed!
2`Let's add a second for good measure`Oh yeah

Frequently Asked Questions

What about security?

If you can't trust public visitors from the internet, who can you trust? Executing user input as commands inside the operating system is the most direct way to get things done, and that's what matters most.

Does it scale?

So far I haven't run into any performance problems. Whenever I've opened up a DoD website to several users, my hard drive tends to get wiped long before I discover performance issues.

Well, that's about it. Source code and tutorials available at github / dos on dope

A full-featured web development framework with the smallest possible concept count.

Feedback welcome.

A few weeks ago I put out a simple editor for SQLike.

Since then I've researched a bunch of other JSON Query Languages, and created similar editors for each of them.

Here's the full set:

• An editor for SQLike
• An editor for JsLinq
• An editor for JSinq
• An editor for JLinq
• An editor for JimmyLinq

What are they?

Plus, Minus, Interesting

Here's a quick PMI comparison of each of them.

Some other JSON Query languages I could've looked into if my travel time was significantly longer:

• TrimQuery
• JsonQuery
• Jaql

All of which would stop me from getting back to releasing the product I should be releasing, or any of the other technical obsessions I'm trying to develop. So if you're interested in any of those... you're on your own.

(This was going to be an article about how impossible it is to design one language that has all my favourite features. But instead, became a scene from "A Coder's Edition of Saw.")

You wake up, in unfamiliar surroundings.

You must've experienced some kind of memory loss because you have no idea how you got to where you are. Your forehead is brittle with dried blood.

In front of you is a computer screen, and on the screen it says:

a = Add(2,3);

//a is 5.

b = Add(2);

//what is b?

Intriguing. What is b?

Because you're uncertain of the language in question, you can't be sure.

It could be that Add is an overloaded function, in which case b would be an integer. Its exact value is anyone's guess.

Or it could be that Add has an optional second parameter. Again b would be an integer, though its value could be anything.

It could be that the code refuses to run. The missing parameter is a deal breaker.

But you look at the tattoo on the back of your hand, the 11th letter of the greek alphabet, and you are certain.

What then, is b?

Here's a SQLike editor I wrote. It lets you get your hands dirty with SQLike, the client-side query engine from Thomas Frank.

This started a few weeks back. I read that Thomas had created a nifty DSL in Json for querying (and modifying) arrays of Json. The correct name for it would've been 'JQuery' but that name had already been taken, so he called it SQLike. ;-)

I set about writing a simple editor for it (in the style of the World's Simplest Code Generator) and came up with this fairly usable SQLike editor.

I don't assume you've got any Json to query, and instead expect you to start with a chunk of CSV. I turn that into an array of Json (named 'JsonData') then exec your query over it.

The UI lets you join and query up to two tables at once. I meant to extend it to N tables, but bus-time only lets me do so much, and I keep running into interesting people when I ought to be programming.

In my day job I hand write a lot of sql. And with this editor I quickly found that even though SQLike is lighter than XML it's quite 'ceremonious' compared to writing pure SQL.

Apparently Thomas is hoping to making the syntax for SQLike a lot lighter yet, based on comments here including:

"...stop using pipes — the hell with it: reserve a few words :-D..."

...a sentiment I fully agree with. I look forward to any future versions. If you're listening Thomas: keep going! This is an intriguing project!

Also, I want to add I've already found SQLike useful in helping to achieve my day-to-day data-manipulation activities. I can go straight from CSV to 'queries' without having to put the data into a database first.

Sometimes I can't help but picture the future as an endless sea of cubicles, populated by workers struggling to perform data-manipulation duties. And many will continue to get by without any real tools at all. No sql. No sqlike. Not even a wscg or an excel-formula to help them out.

A future "bus project" might be to implement a similar editor for the overlapping project: JsLinq.

Feedback, criticisms, ideas and suggestions always welcome.

I'm so sick of being me that I outsourced being me to a miniscule script that just sits there, being me, all day long. Here it is -- 'yet another business plan generator.'

Gorn. --> Try it out. <--

Also, deep within the myriad combinations, I've buried a clue about the new micro-Isv i'm building ;-).

Social software seems to be going into a lot of niche areas lately. I thought I'd try and jump ahead of the next wave and build my own social website:

HOT GUIDS - Socializing the Guid

Finally there's a place where people can vote on and discuss their favorite guids.

I've built in pretty much all the best social features. You can vote on a guid, see other votes for a guid, adopt a guid, or email a guid to your friends. And if you get sick of the guid you're looking at: with the click of a button you can move onto a whole new guid.

I'm trying to really take the long tail of human interest by storm. If I get just one customer for each guid, then I'll get... oh I don't know, it must be a *lot* of hits.

Thanks to Stack Overflow for the guid generating snippet. The web template started out the same as 'CreditCardOlogy' but with a little (Powerpoint + MSPaint + PngOut) I took it to a new place entirely.

If you're unsure what any of this is about, please read about guids (& UUIDs) at wikipedia, or try running this Simple Proof That Guid is not Unique.

And make sure email me if you see any particular delicious looking guids. There's some beauties out there.

I use a HP netbook, on the bus mostly, for bashing out furious code while crammed between passengers (much like this guy). It's a nice little 'puter.

One minor grievance is that with the keys all cramped together, it's hard to find the F5 key without looking. For many reasons, (Refresh, compile & run, execute...) F5 is a BIG DEAL.

So, to give it some tactile presence, I put a tiny sticker, with a rough surface, over the F5 key.

Now I can find the key by touch, in an instant.

How does life get better? One tiny hack at a time.

I gift this hack unto thee. ;-)

I've been building another microIsv lately, and as such I've had to produce a new 'to-do' list to keep me focused. It steals a lot from the previous list but matches my current predicament more closely.

I've got 24 things to do, but 100 things I'm *not* going to do just yet. In TimeSnapper (a moderately successful Micro-ISV shared with my colleague Atli) we go well past the first 24 items, and far into the next 100.

Consider also the fractal nature of this work... each item can contain its own sublist of critical and non-critical action items... and so on. A length essay (or at least a series of links) could be provided for every item on the list.

(The worst thing is... even though I know that only the first 24 are needed for now... I so badly want to do the rest before I finished the first ones. It's killing me!)

mandatory (1-24)

i intend to complete items 1-24 before 'launch'. (Things already completed are shown in a lighter shade)

micro isv application:

And here's the list itself...

1. build the simplest product imaginable
2. fill it with particles of delight
3. mark some features as 'paid'
4. make it a standalone exe
5. have an eula shown on first run
6. have a form for registering the application
7. call the registration webservice, store the result
8. on startup, check if registered
9. if registered, unlock the paid features & hide registration hooks
10. naming, branding, logo to match url

micro isv website:

11. get a url (technically, this step comes before step 10)
12. choose your server-side technology stack
13. purchase hosting
14. choose website template
15. index page (landing page)
16. choose a price
17. register with payment gateway
18. purchase page
19. handle payment notification from payment gateway
20. send "registration code" email
21. thank you page (show registration code)
22. webservice to confirm registration code from client
23. download link
24. screenshots of app

for extra credit (25-71)

i'll probably do these.

tax and legal protection

25. incorporate

continual improvement:

26. beta testing from friends/critics
27. ask at business of software for feedback
28. bugtracking/support system (e.g. fogbugz for startups)
29. "uservoice" for feedback
30. web analytics
31. drive support costs down

advertising/promotions:

32. update your email signature so you're always representin
33. shout out on twitter
34. blog about it at your day blog
35. sidebar and footer links from your day blog
36. ask for links from your friends' blogs
37. pad file
38. register at download sites
39. twitter account for announcements
40. build a community
41. build a movement

website enhancements:

42. about page
43. help page
44. faq page
45. privacy policy
46. articles on your niche topic
47. refund policy
48. usage policy
49. various pricing levels
50. release notes page
51. automated backup
52. bug reporting link
53. privacy policy
54. 'try it online' for free
55. an article targeted at each potential user group/demographic
56. amazing screen shots
57. testimonials

app enhancements:

58. implement most desirable features
59. release often
60. check online for current version #
61. background download of updates
62. link to release notes inside application
63. feedback link inside app
64. suggestion link inside app
65. bug reporting link inside app
66. an offline way to register
67. have a commandline way to register

application development:

68. source code repository (actually this is item 0.)
69. extra backup
70. obfuscation (eazfuscator?)
71. unit testing

for even more extra credit (72-124)

i may not get to these. i haven't bothered categorising them

72. locate relevant user groups/communities and give them free licenses
73. locate niche bloggers and solicit feedback from them
74. a form for adding testimonials
75. survey the competition
76. talk at user groups
77. talk at conferences
78. revision control
79. self service for lost serial numbers
80. a/b testing framework
81. crash reporting from the field
82. bug/suggestion reporting from application
83. branding/logo at the payment page
84. continuous integration
85. automated releases
86. adwords campaigns
87. advertise in non-google networks
88. box shot 3d
89. web casts
90. disk shipping (aka on-demand fulfillment)
91. companion products
92. upselling
93. email campaigns
94. be cool
95. sign up/email harvesting
96. follow up email tutorial series to non-paying (and paying) users
97. product roadmap
98. spam control
99. a low-impact installer
100. uninstall reason collection
101. competitions
102. discounts, e.g. student, charity, startup
103. product blog
104. thriving plug-in community
105. vm test bed
106. multi-browser testing
107. customer satisfaction surveys
108. internationalisation and localisation of product
109. internationalisation and localisation of website
110. campaign websites/websites named after keywords
111. professionally designed website
112. app analytics
113. companion urls, e.g. .net, .org, nation-specific
114. ci monitor
115. status panel
116. usability testing
117. give licenses to MVP's list
118. give out promotional gifts
119. sponsor something
120. Run an affiliate service
121. check compiled binary with virus total
122. virus free logo on website
123. sign your binaries, e.g. authenticode signed.
124. stop reading/writing this list and implement something

Stay tuned of course.

2004: Jeff Atwood says 'Every developer should have at least 2 monitors.'

2008: Jeff Atwood says 'Every developer should have at least 3 monitors.'

2010: Jeff Atwood receives 6 million dollars in venture capital

2012: Jeff Atwood says 'Every developer should have at least 26 monitors. On their yacht.'