Usually I rarely use more than 20GB on my Linux virtual machine. However I set up Dropbox and, once it synced, I had only a few megabytes left on my VM.

These are the steps I’ve followed to resize the VM partition:

Make a backup of your VM

You can potentially destroy your VM so better safe than sorry. Make a copy of your VM.

Delete snapshots

If you have snapshots you must delete them first so changes are merged on the main disk file.

Resize the virtual disk

Run the following command:

VBoxManage.exe modifyhd ubuntu.vdi –-resize 40000

Download GParted Live CD/USB image and use it to boot the VM

You need to download GParted Once downloaded boot your VM with the GParted image.

Resize the partition

If the swap partition happens to be between the primary partition and the unallocated space, you need to remove it. Resize the primary partition and then create the swap partition back.

Be sure to write down the size of the swap partition before delete it and leave that space unallocated when resizing the primary partition.

Once you are happy with changes click apply and exit GParted. Remove it from the boot sequence and restart the VM. Enjoy the extra space.

No soy de los que cambian de móvil cada año. De hecho, el contrato de permanencia que firmé con Movistar hace unos años finalizó en 2011. Y aún seguiría con el mismo iPhone 3GS que adquirí entonces si no fuera porque lo perdí hace 1 mes (historia que no viene al caso).

Desde que lo perdí, he estado usando un iPhone 4 prestado. Comparado con mi 3GS se nota sobre todo la mayor resolución de la pantalla y que funciona algo más rápido. Sin embargo, hay cosas que eran mejor en el 3GS. La batería dura menos, el botón de inicio falla en ocasiones, la señal de WIFI es más débil y sobre todo tiene menos cobertura.

La cobertuta de Movistar no es muy buena en mi zona, ya con el 3GS tenía en ocasiones problemas, pero con el iPhone 4 directamente es imposible trabajar. Prácticamente ninguna llamada me entra directamente y los avisos de llamada llegan tarde. Para hacer una llamada tengo que esperar a ver 3 barras, estando la mayor parte del tiempo con tan solo 2.

Curiosamente la cobertura de Vodafone es excelente en esta zona. Así que decidí cambiar de compañía tras más de 3 años con Movistar.

Aprovechando el cambio de compañía y puesto que el teléfono que tengo ahora no es realmente mío, cogeré el iPhone 5.

Esta ha sido el proceso:

13/02: Hago el pedido en la web de Vodafone. El proceso es bastante claro. Elijo el plan Red Pro y un iPhone 5 32GB. Sale por 59€/mes + 176€ por el teléfono. 24 meses de permanencia. Envío la documentación que piden por email en ese mismo momento para acelerar el proceso.

14/02: Me llaman de Vodafone para formalizar el contrato. El proceso lleva unos 20 minutos pues me vuelven a pedir todos los datos que metí en el pedido. Vuelvo a enviar la documentación, esta vez por fax porque me dicen que suele ser más rápido!! En serio??

19/02: Sin noticias. Por curiosidad me meto en la página de vodafona a ver si hay alguna manera de ver el estado de mi pedido/portabilidad. Si que la hay. Sin embargo no es muy informativa. Decido llamar no sea que haya algún problema. Es un 902 de pago; mala forma de tratar a los clientes. Me atiende una señorita en España, tras pedirme el DNI, me dice que:

Hay algún problema con nuestros sistemas informáticos y no podemos saber el estado de su pedido. Llame dentro de 30 minutos.

¿Cuántas veces habré oído esa excusa? A las 2 horas vuelvo a llamar. De nuevo me atienden de forma local. Esta vez sí me dicen que hay un problema. Por lo visto falta la parte de atrás del DNI, la cual envié junto al resto de documentación. Le pregunto que si no llego a llamar, ¿cuánto tiempo habría estado parado el pedido? En su defensa hay que reconocer que sólo han pasado 4 días laborables.

Vuevo a enviar la documentación.

21/02: Sin cambios en la página de seguimiento del pedido. Llamo de nuevo al 902 y la persona que me atiende me dice que ya disponen de toda la documentación necesaria y se disponen a iniciar el proceso. 4 o 5 días más.

26/02: Vuelvo a llamar tras la falta de información en la página de seguimiento del pedido / portabilidad. Me dicen que ya está iniciado el trámite y que mañana o pasado me debería llamar Movistar para confirmar la portabilidad. Me asegura que como mucho lunes o martes debería estar ya funcionando con Vodafone.

28/02: Parece que ya hay movimiento. Recibo un SMS diciendo que la portabilidad se realizará en la madrugada del día 6 de marzo. La página de seguimiento de Vodafone sin cambios. Más tarde recibo una llamada de Movistar intentando hacerme una contraoferta. Le digo que lo siento pero que es por un problema de falta de cobertura en mi zona. Vodafone me manda un SMS dicicendo que el pedido ha sido enviado. Hay página de seguimiento del pedido. Sin embargo, la página de seguimiento original sigue diciendo que no se ha iniciado ningún trámite.

01/03: Llega el teléfono y la tarjeta SIM. Sólo queda el paso de hacer efectiva la portabilidad.

06/03: Se completa la migración y por fin puedo empezar a usar el móvil con Vodafone.

Resumen

Me ha gustado bastante la página de Vodafone para realizar el pedido. Sin embargo la página de seguimiento del pedido no funciona en absoluto. Deberían deshabilitarla pues lo único que hace es preocupar al cliente y forzarle a llamar.

Tampoco me ha gustado tener que llamar a un 902 de pago para saber el estado de la portabilidad/pedido.

Muy extraño que recomienden mandar la documentación por fax en lugar de por email. Por fax!!

Tres semanas para tramitar todo esto… excesivo!

Por lo menos parece que he solucionado mis problemas de cobertura.

Goal

To have a machine which I can throw ODT files at and get back PDF documents.

(It will work with all OpenOffice supported document types: Word, RTF, PPT, …)

Quick start

Following these steps you should get the conversion working on a clean machine. I’ve used Ubuntu 12.10 minimal server install.

1
2
3
4

sudo apt-get install openoffice.org
sudo apt-get install unoconv
unoconv --listener &
unoconv -f pdf test.odt

Once we get the converter working, let’s wrap this functionality in a rack application using Sinatra.

You can follow this guide, to install RVM, Ruby, passenger and nginx.

Minimal Sinatra app to convert the files to PDF

The following code expects you to POST a file. It saves it in a temp folder, converts it using unoconv and sends you the resulting PDF file.

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20

require 'sinatra'

TMP_PATH = 'tmp/'

post '/' do
 source_file_contents = params[:file][:tempfile].read
 source_filename = get_source_filename(params[:file][:filename])
 output_filename = source_filename + '.pdf'

 File.open(source_filename, 'w') { |f| f.write(source_file_contents) }

 %x[unoconv -f pdf #{source_filename}]
 send_file(output_filename)
end

def get_source_filename(base)
 filename = params[:file][:filename].split('.')[0]
 filename = filename + '_' + Time.now.nsec.to_s
 File.join(TMP_PATH, filename)
end

You will also need a config file for the rack application:

1
2
3
4

require 'sinatra'
require 'converter'

run Sinatra::Application

Now we just need to configure nginx to serve our sinatra app.

1
2
3
4
5
6
7
8

...
  server {
    listen              80;
    server_name         localhost;
    passenger_enabled   on;
    root                /home/jose/converter/public;
    }
...

This is the project folder structure:

1
2
3
4
5

altea/
|-- config.ru
|-- converter.rb
|-- public
`-- tmp

Testing from the command line

Now from other computer you can test the conversor using:

1

curl -X POST -F file=@test.odt http://192.168.3.13 > tmp.pdf

Change the parameters to suit your environment, but that should convert test.odt file to PDF and put the results on tmp.pdf.

What’s next

This is just a proof of concept. The Sinatra app is doing the absolute minimun to get the conversion. It’s not cleaning temporal files for example.

Also the VM is not configured to autolaunch the unoconv listener, nor monitor its status, etc.

Quick answer

You can install .NET 4.0 on Windows XP SP2 machines and run your winforms applications with no apparent problems. Be aware that it is not a supported scenario but if you don’t have any other option you can definitely try it.

tl;dr

Why I’m moving to .NET 4.0

My main product is a Winforms application that runs on .NET 2.0. I’ve resisted the temptation to upgrade to newer framework versions because it’s not easy to ask or convince customers to install additional libraries even if they come from Windows Update- Also, I’ve personally seen many customers still running my app on Windows XP SP2 machines which .NET 4.0 doesn’t support, at least officially.

This app also uses SQL Server 2005 Express as its backend.

Soon, some of my bigger customers will need to upgrade to SQL Server Standard. The easy path would be to install SQL Server 2005 Standard but I don’t think that it’s even possible now. SQL Server 2005 is three major versions behind.

So for those customers is going to be SQL Server 2012. This shouldn’t be a problem as the code should work without changes just fine right?

And it does indeed. Unfortunately, my application uses SMO to create and update the database, to backup and restore data among other fancy things and that didn’t work.

So the first error when I launch my app against SQL Server 2012 is:

 ===================================

 This SQL Server version (11.0) is not supported. (Microsoft.SqlServer.ConnectionInfo)

 ------------------------------
Program Location:

   at Microsoft.SqlServer.Management.Common.ConnectionManager.CheckServerVersion(ServerVersion version)
   at Microsoft.SqlServer.Management.Common.ConnectionManager.InternalConnect(WindowsIdentity impersonatedIdentity)
   at Microsoft.SqlServer.Management.Common.ConnectionManager.Connect()

Makes sense, as I’m using the SQL Server 2005 SMO assemblies. I just need to change those assemblies with the 2012 (v. 11) assemblies and I’m ready to go. But that doesn’t work because the new assemblies use .NET 4.0.

So, finally I have a good justification to move my application from .NET 2.0 to .NET 4.0. I updated all the projects to target .NET4 and run it on my development machine. It crashed with this exception because my app is using third party libraries targeting .NET 2.0.

Mixed mode assembly is built against version 'v2.0.50727' of the runtime 
and cannot be loaded in the 4.0 runtime without additional configuration information.

To solve this I had to add this to my app config file.

1
2
3

<startup useLegacyV2RuntimeActivationPolicy="true">
    <supportedRuntime version="v4.0"/>
</startup>

Installing .NET 4.0 on Windows XP SP2

After installing a Windows XP SP2 virtual machine, I grabbed the .NET 4.0 web installer and fire it up.

It complained about Windows Installer 3.1 not being present. No problem, I installed it and tried again. This time it complained about Windows Image component not being present on the system. I installed that and tried once more:

Downloading…
Installing Client Profile…
Installing .NET 4 Extended…
Installation complete!

No errors.

Finally, the moment of truth. I launched my application and… everything run just fine. I’m still testing it but I’m not experiencing any problems so far.

So, yes, you can install .NET 4 on Windows XP SP2 and run .NET4 applications. At least Winforms applications. Perhaps WPF or ASP.NET applications crash as soon as you run it but for now, my winforms app is doing fine.

The goal is to automatically generate the AssemblyInfo.cs file every time I build my project. I also want to add the current git commit hash as the last part of the assembly version number.

If you spend some time googling around, you’ll find that there are various ways to accomplish this. This is how I do it, step by step:

Install MSBuild Community Tasks

Using nuget this is extremely simple. Open the nuget console and type:

1

Install-Package MSBuildTasks

Edit your project file and add a new target

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25

<PropertyGroup>
    <Major>1</Major>
    <Minor>0</Minor>
    <Build>0</Build>
    <Revision>0</Revision>
</PropertyGroup>

<Target Name="Version">
    <GitVersion LocalPath="$(MSBuildProjectDirectory)">
        <Output TaskParameter="CommitHash" PropertyName="Revision" />
    </GitVersion>
    <AssemblyInfo
        CodeLanguage="CS"
        OutputFile="Properties\AssemblyInfo.cs"
        AssemblyTitle="YourAssmblyName"
        AssemblyDescription="Description"
        AssemblyCompany=""
        AssemblyProduct="Productname"
        AssemblyCopyright="Copyright"
        ComVisible="false"
        CLSCompliant="true"
        AssemblyInformationalVersion="$(Major).$(Minor).$(Build).$(Revision)"
        AssemblyVersion="$(Major).$(Minor).$(Build)"
        AssemblyFileVersion="$(Major).$(Minor).$(Build)" />
</Target>

Update the main Project tag to execute the new Version target before Build.

1
2

<Project ToolsVersion="4.0" DefaultTargets="Version;Build"
    xmlns="http://schemas.microsoft.com/developer/msbuild/2003">

Note that the $(Revision) property, which holds the git commit hash value, is not included in the regular AssemblyVersion or AssemblyFileVersion attributes. The reason is because .NET expects here a number (UInt16) and if we put here a hash it will fail to compile. We use instead the attribute AssemblyInformationalVersion which allows us to put anything we want.

Display the full version in your application

Having access to the exact commit your application is running is invaluable. I display the full version including the commit hash on the About form of my applications. Also, it’s included in the automatic error reports.

1

Application.ProductVersion.ToString();

This summer holidays, I’m going to run an experiment. Could Linux, Ubuntu specifically, become my primary OS?

Currently my primary OS is Windows 7 and I run Linux as a virtual machine using Oracle’s Virtual Box. But, could it be the other way around?

In order to find out I’m going to install Linux on a notebook and force myself to work on it for a few weeks.

I’ve purchased a small (120GB) SSD disk for the notebook because once you have experienced one, a spinning hard drive feels like a 56K modem. Also the notebook is rather old (Dell XPS 1530), so this bump in speed is going to be a must if I want to work with Windows under Virtual Box.

Just to demonstrate than a SSD is the best upgrade you can do to your computer, this is the time it took Ubuntu to boot and let me type my password in both a regular hard disk and in the new SSD:

Seagate Momentus 7200K: 45 s.

OCZ Vertex3 120GB: 17 s.

Worth it!

Installing Ubuntu

I installed Ubuntu 12.04 on this laptop in less than 8 minutes. Awesome!

Then I installed Virtual Box and create a Windows 7 32 bit VM. I gave it 60 GB, half of the disk drive. It’s unlikely that I will need more than 40 GB on Linux, but Windows installs tends to get out of control.

Once I installed Visual Studio and finish configuring Windows to my needs I took a snapshot and put it in a safe place. This way I have a fresh install and I can avoid the tedious of installing everything and in the future.

I will be using this laptop for the most part of August, will see how it goes.

The idea of using Postgres (PG) on my next projects or even migrating some of the existing ones is becoming more and more appealing. I’m already using it in a multi-tenant Rails application and I’m very happy with it.

One of the entreprisy features it offers is replication. Replication is very useful to gain high availability.

I’ve played a bit today with this feature and these are the steps I’ve followed to create a master (read/write) - slave (read) setup on my LAN.

Prerequisites and notes

Be sure to have the same PG version on both machines. I have different versions and was not able to make it work until I upgraded one of them to match the other one.

The data directory using PostgreSQL 9.1 on Ubuntu 12.04 is located in

/var/lib/postgresql/9.1/main/

Master

1
2
3
4
5

wal_level = hot_standby

max_wal_senders = 5
wal_keep_segments = 32
listen_addresses='*'

1

host    replication     all     192.168.3.15/32     trust

Restart the service:

1

sudo service postgres stop

Slave server

Stop the PG service if it is running:

1

sudo service postgres stop

Edit /etc/postgresql/9.1/main/postgresql.conf:

1

hot_standby = on

Create /var/lib/postgresql/9.1/main/recovery.conf:

1
2
3

standby_mode = 'on'
primary_conninfo = 'host=192.168.3.30'
trigger_file = '/tmp/pgsql.trigger'

Final steps

All the configuration files are now ready. The last step is to send a copy of the current master database to the slave server.

1
2
3

sudo -u postgres psql -c ";SELECT pg_start_backup('backup', true)";
rsync -a -v -e ssh /var/lib/postgresql/9.1/main 192.168.3.15:/var/lib/postgresql/9.1/main --exclude postmaster.pid
sudo -u postgres psql -c ";SELECT pg_stop_backup()";

Now you can start the service again on the slave server.

1

sudo service postgres start

You should see something like this in the PG’s log file:

1
2
3
4
5

2012-07-18 12:20:34 CEST LOG:  database system was interrupted; last known up at 2012-07-18 12:19:24 CEST
2012-07-18 12:20:34 CEST LOG:  entering standby mode
2012-07-18 12:20:34 CEST LOG:  redo starts at 0/16000020
2012-07-18 12:20:34 CEST LOG:  record with zero length at 0/160000A0
2012-07-18 12:20:34 CEST LOG:  streaming replication successfully connected to primary

Now you can work with the master database and the changes are automatically replicated to the slave server.

This configuration allows you to use the slave database to perform read-only queries.

If the master server goes down, creating the file /tmp/psql.trigger/ will switch to read-write mode the slave database. Note that the docs say that:

It’s important to prevent the original master from restarting after fail-over, lets you end up with a “split brain” problem and data loss.

I have not tested this fail-over functionality yet.

Useful articles about replication

Binary Replication Tutorial

Streaming Replication in PostgreSQL 9.1

Adventures in scaling PostgreSQL

If you are like me and you are about to write just another command line parser for your little app (or you big app, it doesn’t matter), just stop right there and follow the following steps:

1. Grab this file and include it in you project
2. Read the documentation. In fact, scroll down and read just the sample code.
3. Promise you will never write another command line arguments parser. Ever.

Seriously, Mono.Options from the Mono Project is efficient, elegant and a pleasure to use. And best of all is just one file that you can include in your project, so no need to reference external libraries.

And by the way, if you want to use it with C# 2.0 simply get rid of those lambdas:

1
2
3
4
5
6
7
8
9
10
11
12

// .... code omitted 

bool show_help = false;
List<string> names = new List<string> ();
int repeat = 1;
Options p = new OptionSet();

p.Add("n|name=", "the {NAME} of someone to greet.", delegate(string v) { names.Add(v); });
p.Add("r|repeat=", "the number of {TIMES} to repeat the greeting.", delegate(int v) { repeat = v; });
p.Add("h|help", "show this message and exit", delegate(string v) { show_help = v!=null; });

// code omitted ...

Got some time to play with electronics again. This time I’ve had some fun with a temperature sensor. I’ve improved the tenth sample circuit from the Arduino Starter Kit.

The original circuit/program just reads and transforms the voltage output of a temperature sensor and logs the temperatures to the serial port. I’ve added a couple of leds so when the temperature goes above certain value it lights a red led. If the temperature goes bellow that value then it lights a green led.

I’ve also added a potentiometer so I can adjust the switching temperature value.

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63

//TMP36 Pin Variables
int temperaturePin = 0; //the analog pin the TMP36's Vout (sense) pin is connected to
//the resolution is 10 mV / degree centigrade 
//(500 mV offset) to make negative temperatures an option
int potentiometerPin = 5;

int redLedPin = 9;
int greenLedPin = 13;
boolean tempOk = false;

void setup()
{
    Serial.begin(9600);
    pinMode(redLedPin, OUTPUT);
    pinMode(greenLedPin, OUTPUT);
}

void loop()
{
    float temperature = getVoltage(temperaturePin);
    temperature = (temperature - .5) * 100;

    //to degrees ((volatge - 500mV) times 100)

    float tempSwitch = getVoltage(potentiometerPin);
    tempSwitch = tempSwitch * 40 / 5;

    if(temperature>tempSwitch)
    {
        if(tempOk)
        {
            Serial.println("RED");
            tempOk = false;
            digitalWrite(redLedPin, HIGH);
            digitalWrite(greenLedPin, LOW);
        }
    }
    else
    {
        if(!tempOk)
        {
            Serial.println("GREEN");
            tempOk = true;
            digitalWrite(redLedPin, LOW);
            digitalWrite(greenLedPin, HIGH);
        }
    }

    Serial.println(temperature);
    Serial.println(tempSwitch);
    delay(1000);
}

/*
 * getVoltage() - returns the voltage on the analog input defined by
 * pin
 */
float getVoltage(int pin)
{
    //converting from a 0 to 1023 digital range
    // to 0 to 5 volts (each 1 reading equals ~ 5 millivolts
    return (analogRead(pin) * .004882814);
}

Finally, I connected pins 0 and 1 to a serial RS232 cable to log the temperatures. This way I power the circuit with a battery instead of using the USB port.

I hate when I deploy to production only to discover that the application fails because an error in the assets pipeline, usually because something is not precompiled.

In order to test locally that the asset pipeline is properly set up and to avoid surprises later, I wrote the following script. Before using it, you have to update config/database.yml and add a staging config. The script is self-explanatory:

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58

#!/bin/bash
PID_FILE=tmp/pids/smoketests.pid

prepare_environment() {
    export RAILS_ENV=staging
    cp config/environments/production.rb config/environments/staging.rb
    sed -i -e's/\(config\.serve_static_assets.*= \)false/\1true/g' config/environments/staging.rb
}

launch_server() {
    echo "Launching server..."
    rm log/staging.log
    exec rails s --pid $PID_FILE &>/dev/null &
    sleep 3s
}

kill_server() {
    echo "Killing server..."
    kill -9 `cat $PID_FILE`
}

migrate_db() {
    echo "Migrating staging db..."
    RAILS_ENV=staging bundle exec rake db:migrate > /dev/null
}

compile_assets() {
    echo "Compiling assets..."
    bundle exec rake assets:precompile &> /dev/null
}

make_requests() {
    echo "Requesting webpage..."
    wget http://localhost:3000 --quiet -O /dev/null
    # add more requests here if needed
}

check_errors() {
    errors_found=`cat log/staging.log | grep Error`
    if [ -n "$errors_found" ]; then
        echo -e "\033[31m* Errors found:"
        echo "$errors_found"
    else
        echo -e "\033[32m* No errors! Ready to deploy!"
    fi

    echo -e "\033[0m"
}

prepare_environment
migrate_db
compile_assets

launch_server
make_requests
kill_server

check_errors

prepare_environment() makes a copy of the production environment config and changes the property config.serve_static_assets to true so Webbrick serves static assets.