VIM productivity

Vim Pedal

A couple weeks ago my pinky fingers became so strained it was painful.  This was due to the location of the control key on my keyboard (http://www.kinesis-ergo.com/freestyle.htm).  When typing normally ie. using webrowser, word processor, email etc.  I hadn't noticed.  Control is a modifier key in VIM and used for many commands, requiring me to use it constantly.

Over the last couple weeks a couple of sites had an article about VIM clutch pedals https://github.com/alevchuk/vim-clutch.  I loved the idea but was very reluctant to have to construct it myself, so I searched for prefabbed/complete pedals.  There were a number of programmable pedals including some very nice ones by kinesis.  The issue was the programmable software is for windows.  Once again I did not want this to become a hobby but just something I could buy and start using.

After some searching I discovered Delcom Products Inc. who makes complete programmable clutch pedals.  "Programming" one involves moving pins around to create keyboard key codes, which in turn are just interpreted as keyboard input! Perfect no hassle and no custom software.  I promptly ordered one.  When it arrive it was trivial to open it up and change the key codes to control (included is a huge sheet of key codes and corresponding pin combinations).  The build quiality is very good, the pedal feels very sturdy.  Now I no longer have to stretch my pinkys to reach the control key!



Vim monitor


I am a large proponent of a single monitor setup, instead of relying on hardware for screen space I enjoy relying on software!.  This includes Virtual desktops and keycombinations to place windows on certain parts of the screen.  The main strengths of this are:

1. No bezel! Multiple monitor setups are always split across a bezel! A huge piece of material interrupting the display
2. We can't focus on two monitors at the same time.  If I could make my left eye look at a left monitor and my right eye look at a right monitor then a dual monitor setup would be perfect.  The problem is I can't, there will always be one monitor that cannot be actively used.
3. There is no comfortable multiple monitor setup.  With 2 monitors they can be setup so that there is a "main" monitor directly in front and then there is a monitor off to the side which requires craning the neck.  Or the monitors are perfectly symetrical placed with the bezels touching directly in front of ones eyes.  These means the neck needs to be turned to view either monitor. Three monitor setups "fix" this because one monitor can be placed in front with 1 on either sides.

I have been using a 1920x1200 resolution monitor which has been pretty good.  When I have webiste requirements I split the screen in 2 with the requirements on one side and a browser on the other.  Vim is in a different desktop.  The amount of time it takes to switch from developing in VIM to the browser to view changes (cntr+alt+left) is less time then it would take to move my head to view another monitor.

Just today I tried programming on a 2560x1440 resolution monitor!

My mind was blown.  Normally I can have 4 windows open in VIM and view them perfect and open up to 6 if necessary.  With the higher resolution monitor 6 windows are nothing! Pretty amazing.

Lenovo wins for Linux

Since I switched over to linux as my main OS 2.5 years ago I've had a couple of major hardware conflicts that have kept me from running on certain devices or hardware components.  Linux is maturing greatly.  A wide, ever increasing, range of components are supported.

Luck played a large role in my switching over to linux.  I had an asus g60vx gaming laptop at the time.  Installing ubuntu 10.04 was absolutely hassle free and all hardware was supported by default.  In addition there existed propriety drivers for my nVidia graphics card.  The first install is absolutely crucial in retaining or switching a user to linux (something the community would absolutely benefit from).  Switching is becoming less of an ordeal.  Manufacturers like Lenovo and Dell make it very simple to switch over with full hardware support.

Both Lenovo and Dell have 'Ubuntu Certified' machines.  Any speedbump to installing and running a system smoothly can ruin someones chances of converting.  Individuals do not want to deal with technical issues, especially on their recreation machines.  If Ubuntu didn't offer a list of approved hardware and laptops, I personally would have just gone with OSX and virtualized linux.  Not because I like mac (quite the opposite) but because it is POSIX and it just works with the hardware.  Having to debug, troubleshoot, fiddle with hardware/software all day at work turns me off to doing it at home.  I no longer like spending hours of my small amount of free time debugging linux hardware issues.

There are a couple of reasons I became a Lenovo loyalist:

1.  They offer ubuntu certified machines with compatible hardware.

     The hardware components they choose to use have open source drivers.  When I install fedora or Ubuntu things just work.  Unfortunately there are still issues on messages boards of some certified machines performing poorly.  I've lucked out and have not hand an issue with the T and E series.

2.  Hardware is completely business focused.

    Lenovo hardware is absolutely no frills.  This can be seen in their designs that have barely changed sine the first thinkpads.  Lenovos aren't pretty.  They are functional, reliable, have an extremely high build quality and are military spec.  Plastic is used all around.  But the 'cockpit' is the most comfortable laptop I have ever used.  The keyboard is by far the nicest and most comfortable offered on any portable device.

3. Finally, being conscious of Linux users goes a long way to earn repeat business.

   Most of the large companies offer some sort of linux compatible machines.  Dell has some laptops and HP has many business desktops.  Lenovo goes that extra step and has most of its business/professional laptops certified.


Companies like Lenovo and Dell are helping to lower the barrier to entry for new Linux users.  When linux is not a hassle it increases a user's chance of regularly using linux.

Installing PHP PDO extensions for Drupal 7 on ubuntu 10.04

I ran into some confusion installing php pdo extensions for php5.  Apparently installing has changed significantly with PHP versions over the last couple years.  The catch is PDO extension is installed by default with php5.  To enable it (ubuntu) add the extensions to your php.ini file.

extension=pdo.so

mysql_pdo now comes bundled with php5-mysql package in ubuntu.  So if you are installing on php5 please ignore all the many posts online that suggest installing it through pear, or pecl, if you try these ways you will be very dissapointed.

Once php5-mysql is installed all that is required is enabling it in your php.ini

extension=pdo_mysql.so

restart apache and you should be good to go!!

Django Class Based Views Tutorial -- why I like them!

With the release of Django 1.3 developers can use class based views in addition to the function based views.  Django includes some generic prebuilt views that help to solve some basic problems.  Although they are a great starting point at some point you will probably find yourself having to write a custom class based view.  I hope to clearly explain how that is done.

What are they?

Most of the other functionality in django is already class based: models, forms, middleware.  For this reason views have always kind of stuck out.  They are function based.  urls.py maps to one function.  That function takes a request as a parameter and must return a response object.  Classed based views applies the same oo concepts that are available in other parts of the framework to views.

Why use them?

A common scenario is having a page with a form.  A GET request renders the form and a POST request processes the form.  Although these two processes often share the same url they are very different.  I never have liked cramming these into one function.  

def form(request):
  if request.method == 'GET':
    return render_to_response('the_form_template.html')
  elif request.method == 'POST':
    # do some processing for the form
    # return redirect if successful


Often times I have found myself with some fairly complicated things taking place inside either one of these conditionals.  Code can grow difficult to read, with lots of nested conditionals.  Readability is just a preference of mine and might not bother everyone.   What bothers me the most is these are separate processes that just so happen to share a url.  That does not mean they should share a function though.  Class Based Views solves this problem.  They recognize that these are 2 sides of the same coin but can present them in a more readable way while keeping CRUD operations straight forward.  Class Based Views will automatically call the method associated with the request. Now using new Class Based Views:

from django.views.generic.base import View, TemplateResponseMixin
class TestView(View, TemplateResponseMixin):
  def get(self, request):
    self.template_name = 'the_form_template.html'
    # or you can instantiate your form and render with
    # a template as normal, then there would be no need for
    # the TemplateResponseMixin

  def post(self, request):
    # process the form and redirect.


Additionally, in urls.py there is a small change, instead of putting in the location of the function in patterns such as (r'^yourregex/$', 'project.app.views.function') you import your class


from myproject.myapp.views import TestView
(r'^yourregext/$', TestView.as_view())


If someone makes a request that is not supported django returns a 405 error.


What to use them for?


Whenever I have a url that does different things for different request methods I like to use ClassBased Views. I also use them when I have common tasks to do each request.  For a while I was finding that I had to write a lot of csv download functions, all of them would go something like: 1) validate input, 2)get data, 3) render data as csv.
I found that this was cleanly implemented by creating a generic CSVView.  Every time I had a different download to do I subclassed it and implemented the get_data method appropriately.

For reference django has some great documentation on their Generic Views,  but is lacking in documentation on Class Based Views.  The best source of information on them right now is the source code, easily accessible on github, https://github.com/django/django/blob/master/django/views/generic/base.py

Simple Steps to Problem Solving

For the last couple months, every time I find myself with some downtime I try and answer questions on stackoverflow.com.  I see many questions regarding error logs, and built in functions.  Most of these can be solved by developing a problem solving approach.  There are a couple aspects to this approach.

1)  Read the documentation
99% of the time the best source of information about a library, built in extension, code, etc is the documentation that comes with the code.  I have only very very rarely run into code whose documentation was very poor.  In my experiences I usually skim over documentation, thinking that I can just look at the code samples.  I end up being wrong most of the time.  I can't even remember how many times, I've been stumped and the answers to my questions are in the basic documentation.  Reading every word of the relevant documentation is never a waste of time, after all it's the information the developers believe is necessary to user their code.
 
2) Error logs are not lying
When errors happen they are logged somewhere.  The tricky part then becomes finding out where the logs are.  The other day I was setting up amazon ec2 instance.  I forgot that I had whitelisted the users that were allowed to connect through ssh.   I kept getting a publickey ssh error when trying to connect from my remote machine.  The trick to solving this was to find out where the error log was located ( on the actual ec2 instance) and once I found that out, it clearly indicated that my account was not allowed to connect.  There are so many questions on so involving 500 errors, or php errors.  Knowing where the error logs are and how to read them are an indispensable part of problem solving.
Exceptions are not lying either
In addition to the recommendation above. Errors do not lie.  When the interpreter finds an error on line x.  There is usually an error on line x.  Copying the error and pasting it into google will usually quickly yield a result.

3) Learn to google
Effective google queries is a skill. I wish I had some advice for how to do it.  I think the problem is, the internet has such an abundance of information even a poor googler will eventually find what they are looking for, giving them negative reinforcement.  Learning to effectively google for an answer is a crucial skill.


4) Learn to rely on yourself
Developing good trouble shooting and problem solving skills takes practice.  At first it will be easier to ask other people whenever a roadblock is run into.  But breaking away from this habit will be best in the long run.