The way I see it text messages, call minutes, and all the other limitations when choosing a network all come under one title of ‘data’. At the end of the day a text message should be like an email/iMessage: it simply costs a tiny amount in data to send, irrespective of location across the world. You should be able to walk into a mobile carrier’s shop and be given a list of plans, much like ISP plans; they all have one factor: data. No minutes, no text limits, no extended costs for minutes/texts abroad, just pure data (ideally with sane costs when using data abroad – not the current charge-what-you-want the carriers seem to have gone for).

Phone purchases should be made separately, just like laptops/computers are. Both items are devices with internet-connectivity of some kind; and they should be a separate part of the purchase; no phones would be tied to any specific providers this way (because a phone subsidised by contract is essentially on loan/credit until the 12-18 initial months are over).

The industry needs to split hardware (phones) from service (data). It’s always been this way for computers, and would drive the carriers to more competitive/fair pricing when they can factor out phone cost.

Interestingly over the past few weeks I’v been seeing TV-ads for a new UK network called Giffgaff. Although they list minutes/texts, all their ‘goodybags’ contain unlimited texts, plus a number of minutes and unlimited data. An unheard of offer in the world of the iPhone/500mb limits (o2, three, vodaphone, t-mobile: all guilty of changing unlimited => 500mb in their contracts upon iPhone 4 launch).

/rant over; I’ll be moving to Giffgaff once my o2 contract finally ends.

• crons run without checking to see if an existing one is still running, poorly written scripts will eventually overload the server
• once a cron is complete, it dies and any remaining time before the next run is ‘lost’

My solution to the problem was to write daemons (which are, in most cases, started by a cron). Running a daemon provides a number of advantages/features:

• combined with pcntl_fork, the daemon can dynamically create and stop processes as required
• the daemon periodically retrieves ’jobs’ from the database (with a minimum time between requests)
• each daemon has it’s own ‘lock’ file, to prevent multiple of the same daemon running at once
• a cron attempts to start the daemon regularly, if the old daemon has stopped running, it’s lock file will become stale

An example daemon is listed below, the class requires the PHP pcntl extension (*nix only) & Tudor Barbu’s excellent Thread class →

//daemon (thread func, update func, maxthreads, thread time, update time, lockfile)
$mod_daemon = new mod_daemon( 'thread', 'update', 10, 30, 600, 'test_daemon' );
$mod_daemon->start();

The lockfile is created in the ./tmp directory (eg. lockfile.lock). The update function returns an array of ‘jobs’, where each item is passed as the argument to the thread function. If the last $argv value is ‘force’ the daemon will kill the old process and start a new one.

These daemons must be run from the command line directly, or can always be put into a cron on any timer (with the lock file preventing multiple daemons at once).

You can find the mod_daemon class in this gist →

I’ve never had a custom background on my Twitter before, I guess I’ve never remembered to actually make one. So anyway, here it is. It’s the first time I’ve played around with those PS splatter brushes in a long time.

• an identical apache/mysql server on both machines
• automated backups of all server data
• no problems when there is no internet connection
• avoid carrying portable drives around (they do music/tv/etc)
• avoid having to update git repos on each machine

To keep all the Apache & MySQL data directories in sync, and to also cover the backups, I chose to use Dropbox (after upgrading to the 50gb plan – well worth it). And for the servers, I simply installed MAMP on each machine (plus this wonderful app to stop it prompting for my pasword).

The next step was to setup my Apache & MySQL servers to use the Dropbox (or folders within it) as their data ‘roots’. This is also possible by symlinking folders from within the Dropbox folder. Apache is very simple, in the httpd.conf (on MAMP: /applications/mamp/conf/apache/) file simply change the ‘document_root’:

DocumentRoot "/Users/nick/Dropbox/Fanatical Dev/Webroot"

As you can see, I use ‘Fanatical Dev/Webroot’ as my server directory within Dropbox, but it can be set to any directory. That should mean the basic server is setup (you can see the documents on both servers). The next step is the MySQL server, which is a little more complicated because MySQL keeps log files which we don’t want to sync (they change too often, are 10mb in size and are no use synced). For some strange reason MAMP (not Pro) doesn’t allow you to change the MySQL config file, but still allows you to edit the startup command, so we can set our new settings. On MAMP, the file is located here (on other setups, you’ll need to change the MySQL startup command manually): /applications/mamp/bin/startmysql.sh. It should contain something like this:

# /bin/sh
/Applications/MAMP/Library/bin/mysqld_safe --port=3306 --socket=/Applications/MAMP/tmp/mysql/mysql.sock --lower_case_table_names=0 --pid-file=/Applications/MAMP/tmp/mysql/mysql.pid --log-error=/Applications/MAMP/logs/mysql_error_log --default-storage-engine=MyISAM

All we need to do is add some configuration options onto the end (in bold):

# /bin/sh
/Applications/MAMP/Library/bin/mysqld_safe --port=3306 --socket=/Applications/MAMP/tmp/mysql/mysql.sock --lower_case_table_names=0 --pid-file=/Applications/MAMP/tmp/mysql/mysql.pid --log-error=/Applications/MAMP/logs/mysql_error_log --default-storage-engine=MyISAM --datadir="/Users/Nick/Dropbox/Fanatical Dev/MySQL/" --innodb_log_group_home_dir="/Applications/MAMP/db/mysql/" &

The first setting ‘datadir’ sets the default location for the databases to be stored, again mine is set within my Dropbox folder. The second option ‘innodb_log_group_home_dir’ is the location of the innodb log files (which we don’t want), which I’ve set back to the default data directory MAMP sets.

In addition to the above, my ‘resources’ folder has also been moved into my dropbox. So I now have my perfect setup:

• all server & resource data synced on both machines, and backed up to Dropbox
• if I go away for a weekend, I can just grab my laptop and leave

There are just two this to remember:

• when syncing an out of date machine (ie. laptop after working on desktop for a day) don’t start the server until the sync is up to date
• never run the servers on both machines at once while doing mysql changes

And that’s it!

I’ve cleared away all the old posts and once again, the blog is nice and blank looking.

It’s so true it’s almost unbelievable.

On the plus-side, one of the modules is all about SQL, specifically MySQL which gives me a huge boost considering the number of hours/databases I’ve worked with in the past few years. Some other topics include more Java (oh, joy) programming, hopefully some C and a ‘team project’. This is going to be a killer.

For this team project we are given some kind of software application to build. Knowing the Computer Science staff the application will be something uninteresting and dull, but I still remain hopeful. We won’t be told for another two weeks. The second problem with this team project is that the teams will (most likely) be randomly assigned. The assumption is that we’re all at the same level with each aspect of the course/project, which is bullshit.

The supposed aim of the module is to help us learn how to develop in a working environment. I sure as hell hope I never have to work in a team where meetings are ‘formal’, as in they have minutes and an agenda. Maybe this works in a dull business meeting where numbers are the topic of discussion, but in a development environment you want more open discussions than meeting, where you can bounce ideas of one another.

So it’s looking like another year of nothing-like-real-life computer science.

FD-Core is the more important release, it’s what I use to power my OpenID/FB connect web-apps. I personally believe 99% of apps should be created with OpenID/FB connect as a base; I hate having multiple logins, especially for things such as forums. Why there isn’t a PHP framework focused around this yet puzzles me. I threw together some documentation, except I only got through a few classes before getting bored, check it out here.