If you’re running MAMP, it might help to nuke that too.

Additionally, I recommend getting rid of old config files:
rm /etc/my.cnf
rm /etc/mysql/my.cnf
rm /usr/local/mysql/etc/my.cnf
rm ~/.my.cnf

Comments very appreciated. I made this with GraphViz on my MacBook in a day.

Programming Language Hierarchy v1 (1.2 MB PNG)

1. works with the new Twitter web interface
2. shows you the original text if you click on the translation indicator below a tweet
3. automatically translates any new tweets that show up on a tweet – when scrolling the page, etc

Get it get it!

Pretty simple – it finds any foreign tweets on your Twitter feed and translates them to English, or Spanish, or Japanese, or whatever your native language is. Enjoy!

I am very opinionated about my code.  The source of most of my opinions is the DRY/DIE principal: Don’t Repeat Yourself / Duplication Is Evil.

Take, for instance, a Human Resources application I worked on recently. It is a collection of web forms to get information to and from the HR department at work.  I inherited the system and had to make several improvements: form field validation had to be more robust, submitted form data had to be formatted in a prettier way, fields had to be removed/added, etc.

This was a pain. This application was simple – two forms, less than 50 fields total across each. Total lines of PHP code: 1000 maximum, including comments and HTML. That’s tiny! And yet, within this tiny code base, information was duplicated at least six times. And it was inconsistent – each duplicated piece of information told a different, contradictory story about the web form.

This is why I have internalized the DRY/DIE principal. If you stick with this principal from the get-go, applications are much more robust and easier to maintain. Luckily, my story has a happy ending: I ported 90% of it to PHP-Frame and modified the forms with time to spare, and even got a few nice cosmetic extras in.

tl;dr for this post: next project you’re working on, count how many places you duplicate similar information. Try to minimize that number. I reduced it from 6 to 2 in my application (all the form field validation requirements laid out in the application’s Model; and all the form fields laid out in HTML).  Many methods can be used to avoid (or eliminate, when refactoring) duplicated code and information, including solid framework design (which I have focused on thus far) and using generated code (which I haven’t delved into very much). Regardless of which path you take: don’t duplicate code, and don’t duplicate information.  Duplication is evil.

Further reading:

Rule three [wikipedia]

Single source of truth [wikipedia]

Don’t Repeat Yourself [wikipedia]

Separation of concerns [wikipedia]

I have a lot of standards and ideals with programming, but there are also a lot of gaps. This “best practices” series will be my attempt to explain every facet of my programming philosophy, and hopefully start a conversation.

I think this conversation is necessary because most code we see on a daily basis is bad – real bad. It’s bloated and slow when it works at all. It’s brittle and doesn’t like to change. It probably took forever to get the initial product, but it probably needs a LOT of work – and that’s unfortunate since it’s so difficult to change.

I will probably talk a lot about unit testing and MVC. The reason you should read more is because I don’t like the way those are done now, at all; I think we can do better. The reason I want you to read more is because I know you, as a programmer, are uncomfortable with a lot of what you do. There should be a better way to do things, and I intend to discover it – with your help.

So, before I write my next post, a note : any methods I come up with should be general enough to port to other languages. PHP is what I’m most familiar with, but I don’t intend to be exclusive.

Thanks for reading.

Here:


#!/bin/bash
osascript -e 'tell app "iTunes" to pause' > /dev/null
osascript -e "set Volume 3"
say "Wake up, Josh. Time to get out of bed. You have an exam to take."
sleep 5
osascript -e "set Volume 0.5"
osascript -e 'tell app "iTunes" to play playlist "morning music"' > /dev/null
sleep 5
osascript -e "set Volume 1"
sleep 5
osascript -e "set Volume 1.5"
sleep 5
osascript -e "set Volume 2"
sleep 5
osascript -e "set Volume 2.5"
sleep 5
osascript -e "set Volume 3"


Save it in your ~/bin folder, reference it from your crontab, set up a “Morning music” playlist in iTunes and customize the wakeup message. Also, the tiered volume is fun but really janky.

And my crontab, IN CASE YOU WONDERED:

0 8 * * mon,tue,wed,thu,fri ~/bin/alarm.sh > /dev/null
30 8 * * mon,tue,wed,thu,fri ~/bin/alarm.sh > /dev/null
0 9 * * sat,sun ~/bin/alarm.sh > /dev/null
30 9 * * sat,sun ~/bin/alarm.sh > /dev/null

A snooze feature would be nice, but that can wait.

A lot of PHP frameworks pay lip service to MVC without actually adhering to any strict standards. I see this resulting in (1) duplicated code, (2) duplicated information, and (3) a messy hodge-podge of PHP+HTML, almost every day.

The first symptom (duplicated code) happens because of sloppy programming more than not sticking to MVC. However, the problem is much worse because functional code is placed in the controller, the model, AND the view. Functional code should be placed only in the controller (or “helpers” or “libraries” called from the controller, as the case may be depending on your framework).

The second symptom (duplicated information) occurs because information is placed in the controller and the view. It should only be in the model. For example, form validation: the controller should validate form data, but all validation information should be in the model.

The third symptom occurs because functional code and information is placed in the view. Only the layout and UI design should go into the view. Form validation, sending emails, saving to the database – that’s outside the boundaries of the view, or it should be.

I’m still trying to flesh all these details out. One thing I know for sure that will improve a project by leaps and bounds – don’t allow PHP code in the view! At all. Ever. Some programming philosophies state that, because PHP is a templating language itself, HTML and PHP should be mixed together. I think that’s wrong. It leads to messy, hard to manage, and even dangerous code. My solution is the template engine in PHP-Frame.

RFC 822 compliant email validator for PHP: http://www.iamcal.com/publish/articles/php/parsing_email/

A bunch of handy regular expressions: http://www.roscripts.com/PHP_regular_expressions_examples-136.html
(I’m using the currency regular expressions, but they all look handy)

I don’t have a complete solution yet, but I have a work-in-progress journaling system. If I run a script called “journal-today”, it will open a text file named YYYY.MM.DD (for example, 2010.05.27 for today). It will then insert the time and open the file in VIM for me.

Next, I need a LIST and a THEME feature. The list feature is already working. If I wrap a block of text in <List=”name”></List> tags, my system will display the latest list of that name. If I want to edit the list, it copies the latest version of a list to a new file and then appends the updated list to today’s journal file. Lists are useful for, among other things: todo lists, reading lists, prayer lists, etc. That’s just what I’ve been doing for the past few days.

Next is a THEME feature for related content that doesn’t necessarily need updating.