Cascade of Insights

This post is a work in progress and represents my getting to understand how laziness works in haskell and how it affects performance.

In an old essay, Paul Graham, talks about how some languages are more succinct and therefore they are more powerful.  In fact, some programming languages can say things that you can't easily say in others.  Maybe you can't say them at all.  Its a good essay overall, you should read it.

His Example:

Write a function that takes a number n, and returns a function that takes another number i and returns n incremented by i.

I think this example is, maybe, unfairly biased towards functional programming languages, and its no surprise that pg shows us that lisp has the shortest implementation. 

It works as a good example, however, because it shows us a concrete  demonstration  where one language can easily do something (lisp), that others languages  are not able to do at all (java).

Let me try this in haskell. 

So in Haskell, the type of this function f would be:

f :: Num a => a -> ( a -> a)

meaning f takes a number and returns a function that takes a number and returns another number.

Since the brackets are right associative, we can drop them to get:

f :: Num a => a -> a -> a

This is because, in haskell, all functions are considered curried and can be partially applied.  Even basic numerical functions like + - / if given one argument will return a function that excepts the next argument.

For example:

>:t (+)

(+) :: Num a => a -> a -> a

>:t (+) 4 (+) 4 :: Num a => a -> a

So (+) 4 returns a function that takes a number and then adds 4 for to it. (I mean, I guess technically, it doesn't add 4 to it, but returns a new number that is the sum of them, but I think I am still capturing the spirit of the exercise here.  Haskell variables are not mutable.)

What all of this means, if I haven't already telegraphed the punchline, is that the complete haskell solution is:

f = (+)

So, in haskell, we don't need to write this function. We can just use +. 

How to get bluetooth working in windows 8 / 8.1 on various thinkpads.  

Currently, the thinkpads ( w520 for sure, but lots of complaints about others online) do not have working bluetooth drivers.  Or rather the driver exists, but getting the bluetooth to turn on is not possible.

I have put together a solution here:

https://dl.dropboxusercontent.com/u/7773580/lenovo/Adams%20bt%20enable.zip

This zip has to batch files that will turn on or off the blue tooth device.  It uses the hotkey exe from the windows 7 Lenove hotkey app to force the bluetooth on.

It also contains a bluetooth driver, the one I originally used, however I recommend instead that you go straight to broadcom here:

Problem:

I want to use emacs like I use notepadd++ on windows.

That is:

Right click menu item to "edit with Emacs"

Associate file types with emacs (like .org)

Have only one emacs open at a time

Here are the steps I took:

Create the context menu like this (global_context_emacs.reg) :

REGEDIT4 [HKEY_CLASSES_ROOT\*\shell] [HKEY_CLASSES_ROOT\*\shell\openwemacs] @="&GNU Emacs" # The above value appears in the global context menu, # i.e., when you right click on a file. # (The '&' makes the next character a shortcut.) "Icon"="C:\\Program Files\\Emacs\\bin\\emacs.exe,0" # The above uses the icon of the Emacs exe for the context # and should match the path used for the command below. # The ,0 selects the main icon. [HKEY_CLASSES_ROOT\*\shell\openwemacs\command] @="C:\\Program Files\\Emacs\\bin\\runemacs.exe \"%1\"" # The above has to point to where you install Emacs

I got my new teck keyboard from massdrop. It takes some getting used to, but I’m already typing faster.  Check out the autohotkey script I used to ease the transition.

In my effort to become a better touch typer, I began looking into ergnomic and mechanical keyboards. The cream of the crop for ergonomic symetric mechanical keyboards appears to be the ergodox. It is open source hardware, so you basically have to order the parts your self and get out your soldering gun. They sell these in kit form all the time on massdrop, but this seemed like a little more effort than I wanted to put into a keyboard at this juncture, so when massdrop did a mass buy for the so call Truly Ergonomic Keyboard, which is a compromise between a traditional keyboard and ergodox, I jumped on it and got my TECK as 25% off.

Keyboard arrived very quickly from massdrop and I've been using it full time for about a month now. I made some modifications to the keymapping using the the included software, switching the control and shift keys to more traditional spots. I also used AutoHotkey on windows and xcape on Ubuntu to get my right control to work as an enter key if pressed alone and some other similar tweaks.

Notes from first week of use.

Having the shift keys on home row is very smart

but i want to be backward compatible with traditional keyboards so switched it back

enter is middle is great

delete and backspace take getting used to

quotes key in new place is great for typing text, but not good with VIM and the shell

Backwards compatible tweaks

I want to be able to use normal keyboards when I have to, so I made some tweaks to the TECK keyboard to keep it close to default layout.

switched key caps - CTRL and SHIFT

right control is enter if pressed alone

left control is ESC alone (vimness) means my control is like my old caps lock

tried mapping shift on its own to tab, but messed up my shift tabbing

References:

My Autohotkey scripts  for the TECK

TECK website

How many Sundays fell on the first of the month during the twentieth century (1 Jan 1901 to 31 Dec 2000)?

http://projecteuler.net/problem=19

By Hand:

There are 12 * 100 first of the months, 1/7 of them fall on sundays, therefore euler19 = 1200/7 ~= 171

In haskell:

What is the greatest product of four adjacent numbers in any direction (up, down, left, right, or diagonally) in the 2020 grid?

http://projecteuler.net/problem=11

It took me forever to find out that transpose existed and that I could use zipWith to get a diagonal - but after that its gets easy.

Using the code from the preceding three exercises, implement Graham's scan algorithm for the convex hull of a set of 2D points. You can find good description of what a convex hull. is, and how the Graham scan algorithm should work, on Wikipedia

http://book.realworldhaskell.org/read/defining-types-streamlining-functions.html

I am reading "Joy Of Clojure".  It seems like a good book, but it keeps telling me about these problems I have which I don't percieve as problems.

I'm not looking for Clojure to save me from these "problems" I just started playing with it because of some Rich Hickey talks that made me think there was a better way to develop software.

Problems:

Order of operations

Prefix notation is simpler, but order of operations have never been a big problem in my programming life.  

The Expression Problem

http://en.wikipedia.org/wiki/Expression_problem

Protocols seem cool.  But C# has extension methods, so adding a method to a class after the fact is no problem and the namespace scoping of extension methods avoids most monkey patching problems

Concurrency

Let's say that you won the lottery. That would be great, wouldn't it?

Now you're rich because of this lottery ticket and i ask you: was a good idea for you to buy the ticket?

Answer is no. No it was not a good idea for you to buy that ticket. The expected value of a lottery ticket is negative.  It is a waste of money!

Lottery tickets are a bad investment.  Even though you won that does not mean it was a good idea for you to get the lottery ticket. This is an example of the creeping determinism.  It is related to Hindsight bias, but it is distinct.  

“the sense that grows upon us, in retrospect, that what has happened was actually inevitable”

You were not destined to win the lottery.  There is no fate.  You took a gamble, but just because it pays off does not mean that it was good gamble.

The Generic Case:

Just because something happened, doesn't mean it was destined to happen.  Life involves a lot of random elements. Just because things went this way does not mean that they were destined to go that way.

Creeping determinism means that we pay less attention than we should to the things that don’t happen. But we also pay too little attention to most of what does happen. We notice when we just miss the train, but not all the times when it arrives shortly after we do.

This is covered in the book "everything is obvious" as well as by Gladwell in the new yorker.  They all have much better examples than I.  Also this was dictated into an Iphone, so hopefully it makes sense.

Force categories list not to use category descritption use this:

  $args_list = array(

'use_desc_for_title' => 0

  );    

Code Generators, You Suck.  Well, some of you do. 

The scaffolding style code generators that generate out a good starting point from which standard coding takes over, those are very useful.

The Microsoft brand "design surface" based code generators (TableAdapters,WSSF) where the uml diagram of a class needs to be edited, rather than the actual class, those are more harm than good.  if fact they slow development way down.

    public static class ExceptionHandler

  {

    [Serializable]     public class ReturnOnException : OnExceptionAspect     {         public ReturnOnException(object objectToReturn)         {             _objectToReturn = objectToReturn;         }         object _objectToReturn;         public override void OnException(MethodExecutionEventArgs eventArgs)         {             string message = eventArgs.Exception.Message;             eventArgs.FlowBehavior = FlowBehavior.Return;             eventArgs.ReturnValue = _objectToReturn;         }     }

}

Usage:

//returns -1 on any error

[ExceptionHandler.ReturnOnException(-1)]

You have made a project using attribute based caching, but the cache isn't being hit?

Does the test project included in the source work?

if you look at the test.csproj you can see the postsharp targets line: http://cache.codeplex.com/SourceControl/changeset/view/5279#75842

<Import Project="..\Cache\lib\PostSharp\PostSharp.targets" />

You need that line in project containing the attributes or alternatively you can run the post-sharp installer which will handle this automatically.

Generic Hierarchical Meta data is cool!  There are many ways to implement it with a standard relational database.  Most of them are considered anti-patterns, but sometimes we need to optimize for flexibility, not raw database performance.  Here are some of my notes on the subject:

If all entities in the system can contain a list of tags, and the tagging system is structured correctly, the following features can be supported:

faceted search

hierarchical organization (categories, sub-cats and so on represented in tag hierarchy)

ordering entities ('high priority tag")

This should allow for a flexible meta-data system over top of the entities which can be customized with very little code modifications.

This can be accomplished by having a tag tree, where tag entities have the following fields:

Name

ParentId

Weight

Hierarchy Path (for sub-tree searches)

Shortcut

The following tag search can then be implemented in SQL:

And search (tag X and tag y)

Or Search

Except search

Sub-tree inclusive or not

Assuming that most tag trees are only 2 levels deep and using a subtree inclusive search when this is not the case (or always), we treat each tag tree as a facet in a faceted search. 

A hierarchy of categories and sub categories can be built by building the tag hierarchy and assigning all items into leaf tags and once again using an inclusive tag search as we drill down.

If tags in a tree are given weights, entities can be sorted by these tags in , for instance, a grid. 

Tag Input:

specify path example: facet1/tag1, color/red

User shortcuts to aid entering new tag {name="color",shortcut="#c"}, then can be used like #c/red

Other considerations:

Null Tag :

Null tag could be added to any subtree which would catch all entities without a tag in that tree.  This would be helpful for facet navigation and search. 

Null tag for facet:

A null tag of "Unknown color" could be added as a child of the color tag, and would return all entities without a color tag.

Null tag for sorting:

A null tag of "no priority" could be added as a child of the priority tag and be given a weight of zero, hence sorting on this priority tag tree would return un-prioritized entities last.

Need cohesive example. Another Post.

Tagging Links:

How to Make a Faceted Classification and Put It On the Web | Miskatonic University Press

Faceted classification - Wikipedia, the free encyclopedia

Collection: Faceted Navigation

Advanced Tagging — Hierarchical And Ordered Tags « PubNotes

Advanced Tagging - Things Wiki

Tags as categories:

http://www.znode.com/