Popular Posts

Tuesday, March 29, 2011

State of the Open Source Movement

The open-source scene pretty much took the fun out of programming. It's become an industry of egomaniacs engaged in constant one-upsmanship. There is little-to-no collaboration, constant reinvention of the wheel, and little done of actual technical merit -- most work tends to be the manipulation of known and well-documented APIs, or the debugging of someone else's code. All in all, it has become distasteful.

There is a lot of grief for little, if any, reward. Open source users -- not to mention developers -- are more impatient, demanding, rude, and insulting than their counterparts in the commercial world. Why put up with this when people will pay you for your time and effort, and actually thank you for taking the time to read a bug report?

You know, I was reading the Van Roy paper and took a nap. I had this dream where I encountered this really nasty bug in something I released, so I stayed up late figuring out how to fix it, uploaded a new version, then posted a news item telling users not to use the previous version, and what the (now fixed and closed) bug was. This post got a few comments saying stuff like good catch, validated that the new version does indeed fix the bug, etc.

Now I woke up a bit later and thought all over this and realized that this is completely the opposite of how open source projects work. Commercial software works this way, but not open source stuff. The lesson here is that you don't want IT people as your userbase; they're bastards; even though they should be more understanding than any other people that software takes time, bugs happen, and some problems aren't solvable.

I think it goes back to the social problem of many IT people- every coder thinks they are better than other coders, even when their stuff sucks. Coders look down on sysadmins (who are often their users), and sysadmins in turn think that coders don't know their jobs as well have the deluded belief that writing a large multithreaded highly complex piece of software is as easy as using a scripting language to read text logs.

So yeah, to be happy, write software for closed source.

You know, since that stupid bazaar vs cathedral essay came out, I always got the impression that the cathedral was a nice, serene place of thought and meditation, and that the bazaar was a place one would get stabbed during a shouting match over haggling about some feature/bugfix.

Sunday, March 27, 2011


Researchers found that a small minority of users-- around 0.05% of the site's population-- are generating half of all Twitter posts.

I can see the plausibility of using something like facebook for networking, for some lines of work [sales etc]... but twitter has no observable utility, all those feeds would be better served by an RSS feed.

I don't want to be notified, say, when the beer changes at Toronado, I want to know what is on tap tonight.

Sending from SMS means that any message is not going to be well-thought-out and will likely be vague or incomplete, and receiving via SMS means constantly getting interrupted by ill-conceived messages. It just sounds like lose-lose to me; the nature of Twitter guarantees that the posts are going to be inane drivel. With RSS there was a chance you'd get useful info (though Google news turned out to be a better way of getting headlines). With Twitter feeds, there seems to be no chance of getting useful info; it's the nature of the medium.

I'm not saying RSS is great or that I rely on it; but I fail to see how Twitter is in any way an improvement (for the consumer of the feeds i mean... obviously sending via SMS is convenient for the publisher). They are useless for the consumer; for the producer, sure, but who cares if it is convenient for the producer? They are the ones making money off the thing!

Thursday, March 24, 2011

the rest

markup langs:



go do the xml thing first
1. http://www.w3schools.com/xhtml/xhtml_intro.asp
2. http://www.w3schools.com/xsl/xsl_languages.asp
3. http://www.w3schools.com/xpath/default.asp
4. http://www.w3schools.com/xslfo/default.asp

5. sgml



So far the documentation is anoying; every time I read a document it has jargon that is either java PL buzzwords or scala specific.

The most important things in scala seem to be:
pattern matching
first class functions
case classes
- http://www.scala-lang.org/node/107
- http://webcache.googleusercontent.com/search?q=cache:45KFeYN40gEJ:www.codecommit.com/blog/scala/case-classes-are-cool+what+are+case+classes&hl=en&client=firefox-a&gl=us&strip=1

new operator syntax for embedded dsls
type inference
- http://patricklogan.blogspot.com/2007/06/scala-implicits.html
structural (duck) typing
import syntax
- http://www.scala-lang.org/node/242
- http://www.scala-lang.org/node/126
problems with scala:


dumbing down the programmer:


DSL. DSL. DSL! http://en.wikipedia.org/wiki/Domain-specific_language
Think in terms of DSLS and blocks. Everything in terms of blocks and defining
a langugage to fit the problem!


Tuesday, March 22, 2011


All languages use libraries, so that isn’t really quite a distinguishing feature. However you should probably pay attention to the *quality* of libraries available.
Java and Python have the best general libraries, hands down. C++ is good with QT.

http://www2.research.att.com/~bs/bs_faq2.html (bjarne’s style/idioms faq)

- qt smart ptrs:

PIMPL (pointer to implementation idiom)
http://www.gotw.ca/gotw/028.htm (fast pimpl)




Look at the syllabi:
  1. Introduction to the STL
    • purpose and structure of the STL: containers, algorithms, and iterators
    • review: class and function templates
    • container classification, characteristics, and content
    • iterator classification and behavior
    • generic algorithms and iterators
    • design of generic algorithms and performance guarantees
    • iostream iterators
  2. Generic Algorithms
    • review: generic algorithms and helper templates
    • interaction between algorithms and iterators
    • generic algorithm goals, documentation, and design
    • sequences, subsequences, and sequence errors
    • descriptions of STL algorithms
  3. Containers
    • properties, insertion and deletion effects, specific functionality
    • vectors
    • lists
    • maps
    • containers and exceptions
    • choosing an appropriate container
    • container adapters
  4. Function Objects
    • functions and generic algorithms
    • function objects and generic algorithms
    • algorithm families
    • function objects and containers, strict weak ordering
    • standard function objects
  1. Adapters
    • function object adapters
    • binders and negaters
    • helper tempates and template argument deduction
    • pointer to function adapters
    • composing new function objects
    • designing function object adapters
    • container adapters
    • priority_queues and heaps
  2. Iterator Design
    • the iterator base class and iterator traits
    • review: class template specialization and partial specialization
    • iterators and const_iterators
    • accessing iterator information in generic algorithms
    • versioning of generic algorithms based on iterator class
    • reverse iterators and iterator adapters
    • insert iterators, adapters, and helpers
    • raw storage iterators, adapters, and helpers
    • designing new iterators
  3. Container Design
    • containers vs. STL containers
    • STL container requirements
    • design of a simple STL container
    • design of a more complex STL container
    • containers, iterators, and const_iterators
    • basic STL components
  4. Allocator Design
    • containers and memory management
    • allocators and the standard allocator
    • allocator portability problems
    • container use of allocators
    • design of a special-purpose allocator
    • allocator hacking techniques
"i'm used to writing async code already and this is all neatly tied to the posix layer, which makes non-blocking network stuff ridiculously easy. there's also the package system (npm), async test suites, no worries about messing up the global namespace, etc. you gotta OOP tho if you just throw literals around you can't have or protect a local scope"

Javascript (node.js) http://nodejs.org/ review:


  • downside, requires libssl-dev (fixed with -no-ssl)
  • install with “git clone https://github.com/joyent/node” (ryan is now joyent)

Use Prototype Programming, Delegation.
prototype based programming -http://coweb.cc.gatech.edu/cs2340/4288
// Example of true prototypal inheritance style in JavaScript.
// "ex nihilo" object creation employing the literal object notation {}.
var foo = {one: 1, two: 2};
// Another "ex nihilo" object.
var bar = {three: 3};
// Gecko and Webkit JavaScript engines can directly
// manipulate the internal prototype link.
// For the sake of simplicity, let's just pretend
// that the following line works regardless of the
// engine used:
bar.__proto__ = foo; // foo is now the parent of bar.
// If we try to access foo's properties from bar
// from now on, we'll succeed.
bar.one // Resolves to 1.
// The child object's properties are also accessible.
bar.three // Resolves to 3.
Don’t bother with Normal OOP. tho you can :
  • http://msdn.microsoft.com/en-us/magazine/cc163419.aspx
  • http://en.wikipedia.org/wiki/Prototype_based_programming
  • delegation http://en.wikipedia.org/wiki/Delegation_%28programming%29
  • http://www.w3schools.com/JS/js_intro.asp (javascript)
  • http://nodejs.org/docs/v0.4.1/api/all.html#__filename (node api)
operator overloading (not supported, this spec is fake)

node examples:
https://github.com/nodejitsu/nexpect/blob/master/demo.js (uses spawned procs)

“DOn’t bother with OOP, just learn the DOM”

look at the GOOG custom search API, it’s in JSON:

Monday, March 21, 2011

Programming Languages as of March 2011

Tiobe's chart which force me to take stock of my favorite programming languages. Erlang is #39. It's pretty obvious it's nothing more than a POC language to take ideas from. Scala is even worse, down at #44. Go is at #14 and falling, and pretty much a one trick pony (Go routines).

Sunday, March 20, 2011


The “ESR 3”


Strain’s Map:

Thursday, March 10, 2011



ふい (coming to nothing, ending without result, total waste)

Saturday, March 5, 2011

How to Avoid Stupid Startup Recruiters

"So. What do you do?"

This is a question i hear more and more in the dot com cafe section of District Five - the cafe where all the hipsters fight over power plugs so that they can run their latest "XML and Java to put Stupid Stuff on the Web"(tm) startup from when the cafe opens at 6am to when it closes at midnight.

It is a question that makes my skin crawl.

I usually reply, "I am a programmer." If they still look happy to hear that, I continue with, "A programmer is a professional coder, which just means that you have grown the hell up and you have a freaking job and and you don't care about stupid stuff on the web cos you're making money and are acting as an ADULT.

When they try to tell me about all the wonderful web 2.0 bullcrap, I point out that I program things for medical barcode readers, and other medical software. "The stuff that keeps people alive."

Most piss off about then.

If they bother me more, I tell them that when I'm not at work, I try to stop talking about work.

Wednesday, March 2, 2011

Why I don't Use Facebook

People keep asking me when I will use Facebook.

Here's the long version of why I do not. Every time we choose to use a product we are choosing to support a design decision. Actually several design decisions, as engineers out there will know, since every project is a conglomeration of tradeoffs. The one major thing I cannot support in good conscience, is PHP.

Let me just give you a few quotes from the man who designed, maintains, and gave us PHP:

There are people who actually like programming. I don't understand why they like programming. -Rasmus Lerdorf

So as you can see, clearly, programmer happiness isn't something that Mr. Lerdorf cares about. So you can expect that since this isn't a priority, that coding in his language will be full of Pain and Suffering.

I don't know how to stop it, there was never any intent to write a programming language [...] I have absolutely no idea how to write a programming language, I just kept adding the next logical step on the way

I know how to stop it, Mr. Lerdorf. And it is called the Free Market. If everyone stops using a product written in PHP, then there will be no more engineers demanded who write PHP. Then there will be no more PHP.

And so that is why I will not use Facebook.