e.g. /etc/NetworkManager/system-connections/GrammeenPhone

[connection]
id=GrameenPhone
uuid=b68882db-7e97-4fa3-a811-9234f64098f6
type=gsm
autoconnect=false
timestamp=1316411329

[ipv4]
method=auto

[serial]
baud=115200

[gsm]
number=*99#
apn=gpinternet

#nmcli con up id GrameenPhone

Last year I had a motorcycle accident and broke my 3rd metacarpal in my right hand.  At the same time my power cable for my laptop went out and getting a warranty claim processed in Bangladesh takes longer than it should. I was without my primary computer, and I couldn’t use a keyboard like I normally did. The biggest single reason I bought an Android device is that it works as a standalone device, and during that time, that functionality served me well. For several weeks, my Nexus One became my primary computer. I used it one handed, and it turned out that for many things I wasn’t really limited. There was a big criticism about tablets was consume content rather than create content. I think has largely faded being I don’t read as much about that anymore.  High end phones are just as capable processing wise as the tablets and maybe a more important question is whether the device functionally complete: is the device able complete a task without physical access to another general purpose computer. While my Nexus One was my primary device, I couldn’t access peripherals such as flash storage and USB devices and I couldn’t develop on the Android for the Android.¹ These issues are basically the same on iOS as well, but the times they are a changing.  Smart phones are becoming increasingly standalone devices.  I’ve now got a Python environment on Android through Scripting Library for Android and the recent announcement of Android ADK is going to allow a plethora of hardware options.  Apple with it’s proprietary connecter offers a good deal of input output options and dongles like camera connection kit allow you to move data on and of a device. Recently I even saw that Oscium makes an oscilloscope/logic analyzer for iOS, which is something I would find useful.

As these phones become functionally complete, they’re going to start replacing computers.  As low end phones gain functionality and processing power, Android is going to start showing up in the developing world.  I wonder if Microsoft realizes how truly irrelevant it is becoming.  Desktops and laptops here run either open source or pirated software, but there are over a billion people who want to buy phones in the Indian subcontinent and another billion in China.  While I don’t think Apple is going to go after the developing world market, I suspect that Apple will continue to transition iOS toward a standalone OS which would position them with the option.² The personal computer of the developing world is a phone made in Shenzhen, and I wonder who’s thinking about that.

1 Not including a linux environment on a rooted handset.
2 My bet for the North Carolina Datacenter.

It’s starting to get hot and as a result power outages are getting more frequent.  Last year with more than 12hour daily power outages and only 3 or 4 during daylight hours it was hard to get any work done.  This year however, we got a solar system setup and I hope that makes things go smoother.  There are still some quirks to work out, but we seem to have a system that is giving us consistent power through the day.  We have 960W array and a 48V 400ah battery pack.  The interesting bit is the charge controller, a Su-kam PCU 2k48.  I haven’t seen anything quite like it and it seems like a fairly ideal system for our situation.  There’s no net metering policy in Bangladesh, and even if there was, the grid power is so inconsistant that you would need some sort of battery backup anyway.   This unit ties together PV, grid and a batteries.  I haven’t tested inverter efficiency yet, but the design seems like about the most sane way to do it.  The documentation is fairly confusing, but there are two paths though it. There’s the PV -> Charge controller/Battery->  Inverter path which it prefers, but in the event that it can’t supply enough from that source, it fails over to the other path which is Grid power, also there is a battery charger which can charge your pack at the same time.

I spent another day in Kaula Lumpur, however that was mostly a non event.  I stayed out at LCCT and while I did go in to the town center, I didn’t see much.  This morning I camped out in a Starbucks taking advantage of their free wifi.  I’ve been thinking about a redesign of the old blog for a while now, something a little more minimalist and this seemed like a good way to use a slow morning.  I’ve been reading Dive into mark and I’m liking a lot of his thoughts, especially things like Simplicity is hard lets go shopping.  So I started hacking this morning and by check in time I had something that looks like his site, maybe too much, even if it is the MIT license.   I’m not sure how important blog design is these days, I mean I use RSS feeds for most of my friends blogs.  It’s fun to have something that looks good though, or at least that I think looks good.  If it doesn’t look good, well, stop using Microsoft explorer, or let me know.

#!/usr/bin/python
"""
twitterd
Author: Nathan Charles <ncharles at gmail dot com>
Version: 0.1
This program twits the contents of a file when it's time stamp is updated.  
The functionality is similar to that of finger/.plan in days of yore.
Requires:
   pytwitter
   configobj
This program has no warranty to the full extent of the law
"""

import pytwitter
from sys import exit, argv
from os import path
from time import sleep

POLLINT = 60
READLEN = 140

def setEnv():
    from configobj import ConfigObj
    from os import environ
    import getopt
    env = {}
    HOME = environ.get("HOME")
    config = ConfigObj(HOME + '/.twitrc')
    env["watchfile"] = HOME + '/.plan'
    try:
        env["username"] = config['username']
        env["password"] = config['password']
    except:
        #.twitrc doesn't exist not nessarily a problem
        pass

    try:
        env["optlist"], args = getopt.getopt(argv[1:], 'dfh',["username=","password="])
    except getopt.GetoptError:
        usage()
        print "called exception"
        exit(1)
    for o,a in env["optlist"]:
        if o == "--username":
            env["username"]=a
        if o == "--password":
            env["password"]=a

    if not "username" in env:
        print "username is not set"
        exit(1)
    if not "password" in env:
        print "password is not set"
        exit(1)
    return env

def watch(username, password, watchfile):
    """Updates a twitter feed with the contents of a file when the timestamp is updated.
    """
    try:
        timeStampInitial = path.getmtime(watchfile)
    except:
        #file doesn't exist setting to 0 because it might exist in the future
        timeStampInitial = 0
    while 1:
        try:
            timeStampIncrimented = path.getmtime(watchfile)
        except:
            #file still doesn't exist
            timeStampIncrimented = 0
        if (timeStampInitial==timeStampIncrimented):
            sleep(POLLINT)
        else:
            timeStampInitial = timeStampIncrimented
            #update status
            statusTxt = file(watchfile).read(READLEN)
            client = pytwitter.pytwitter(username=username, password=password)
            rc = client.statuses_update(status=statusTxt)
            print rc

def usage():
    """Prints when called with no arguments or with invalid arguments
    """
    print """usage: twitterd [options]
   -d    daemonize
   -f    run in forground
   -h    help

   twitterd must have twitter username/password set. this can be set via .twitrc

   --password    set username
   --username   set password  
    """

if __name__ == "__main__":
    g = setEnv()

    if g["optlist"]:
        for o,a in g["optlist"]:
            if o == '-h':
                usage()
                exit(1)
            if o == '-d':
                #run in background
                import daemon
                daemon.daemonize()
            if o == '-f':
                #run in forground
                print "Running in Foreground"
    else:
        usage()
        exit(1)

    watch(g["username"], g["password"], g["watchfile"])

Arduino Thermistor Sketch

Then I wrote some quick and dirty python code to read the values coming off the serial line and write it to a csv file which I could open with OpenOffice.

import serial
import time
ser = serial.Serial('/dev/tty.usbserial-A6008dxP', 9600, timeout=1)
logfile = open('test.csv', 'a')

while 1:   # read a '\n' terminated line 
   line = ser.readline()   # read a '\n' terminated line
   if not line:
      break
   words = line.split()
   now = time.strftime("%d/%m/%Y %H:%M:%S", time.localtime())
   a =  "%s, %s" % (now, line)
   if line.find(',') != -1:
        logfile.write(a)
        logfile.flush()

logfile.close()
ser.close()

The resolution isn’t that great because the inputs are only a 10bit ADCs but it’s more than enough to get trends. Without a more accurate thermometer, I can’t tell for certain how accurate it is, but as far as I can tell it reads faster and more accurately than the mercury thermometer we have. In university I would have done this with a fairly expensive LabVIEW system. The impressive thing is that I can get good enough information for what I need with Python and less than $30US in parts including the Arduino.