Happy coding

Yesterday’ve been playing around with CFS in my Ubuntu. I’ve never tried a cryptographic filesystem before, altough I knew there were many flavours laying out there.

CFS has two main advantages, as far as I know:

• It doesn’t require any special kernel patch (at least, not for Ubuntu), because it uses NFS to do loopback mounting of crypted directories.
• It uses directly the underlying filesystem, avoiding the need of creating fs images to mount by loopback. Crypted directories and files are mapped to normal directories and files with its name and contents crypted.

The package provides some utilities:

• cmkdir: Creates and initializes a crypted directory on the host filesystem.
• cattach: Attaches a crypted directory, making it available (as cleartext) under /crypt/*.
• cdetach: Dettaches a previously mounted crypted directory.

I’ve tried CFS successfully, and added it to my list of useful tools. It’s clean, easy and powerful. 🙂

While helping Xavi with a small task (batch converting ODT documents to PDF via a web form), I’ve remembered a little recipe I’ve written some time ago.

It’s a way to force some script to die after a defined timeout. So, you can perform a command by ssh inside a script or perform any blocking task you want, and it will always return.

The only thing you have to do is to put these single 3 lines of code at the start of your bash script:

SECURITY_TIMEOUT=60
CMD_PID="$$";
{ sleep $SECURITY_TIMEOUT; kill -9 $CMD_PID; } 2>/dev/null &
Enjoy it!

Some months ago, I configured a complete mail gateway and other services for a client. As for almost every deployment, there were things that had to be investigated and one always learn something new.

One of the most surprising “curiosities” I found was about SMTP. When I telneted the SMTP gateway at port 25 from inside the intranet or localhost, the usual welcome message was displayed. But when I did the same from the internet, I received this weird welcome string:

220 **************************************

And when I tried to issue an ESMTP command (like EHLO), the server said that it wasn’t supported. What the hell??!! It can’t be! Someone or something is changing my packets!

After doing some searching at Google, I discovered who was the little guilty: a Cisco PIX firewall configured with the “fixup protocol smtp 25” option turned on. That was preventing internet users from authenticating and using TLS.

Thank you debian-administration.org guys!

Past week I’ve travelled to Valencia to take a rest and see the city. A big city, by the way, altough people’s timetable was a bit “strange”, but that’s a different story…

One of the most surprising things happened in the flight: in the plane there were glass coasters, napkins and even headrest covers with the advertisment “sé legal, copia GNU/Linex” (“be legal, copy GNU/Linex”).

It seems like Junta de Extremadura has made an agreement with Iberia to show Linex advertisments and offer Linex CDs and leaflets in some regional flights. What a good thing!

PhpReport is a GPL web application that tracks tasks performed by a company staff or workgroup and allows administrators to launch queries and get statistics about the tasks and time spent on each one. This give company managers the proper information to do cost analysis and decission making. PhpReport is used internally by Igalia staff on daily basis since December 2003.

This summer Igalia has planned a “company practices” hiring to improve PhpReport. Jesús Pérez is the student selected, and he will make use of XP Tracker to manage his tasks. At this link you can follow his progress.

The improvements assigned to this development iteration satisfy new functional needs and publishing & community issues. Some of those improvements are:

• Internationalization.
• Migration of existing OpenOffice.org spreadsheets into PhpReport, to make result extraction and business decission making easier.
• XML import module rewriting, to improve mantainability and extensibility of the application.
• Administration and usage documentation.
• Package improvements.

As a result of this iteration, a new release is planned for the end of the summer. Stay tuned!

I’ve worked a little more in NeoSudoku and at last it can:

• Solve human targeted sudokus using the limited resources that the cell phone has.
• Learn new sudokus (new board configurations input by hand)
• It can’t save new sudokus to persistent storage, but it will be the next feature to come.

The rest of the features can be found on my past post. You can download it from here:

• Source code: NeoSudoku-1.2.1-gpl.zip
• Jad descriptive file: /neosudoku/NeoSudoku.jad
• Jar game file: /neosudoku/NeoSudoku.jar

Enjoy it!

This is the current J2ME/MIDP project I’ve been working on at my spare time since the past weekend: An implementation of the known game Sudoku.

It’s not published yet, because I want to enhance it a bit more and at this moment it contains some sudokus borrowed from a copyrighted book, so publishing wouldn’t be a good idea…
Currently the project is at version 1.1, having the following features:

• Complete implementation of the game rules, with detection of illegal number placement and end of game. The game is playable.
• Several pre-stored sudokus can be chosen for playing.
• Special hints available:
• List of available values that can be put into a square.
• Board map showing the number of values eligible to be put on each square (good to detect the squares that can only be filled with one single value).
• Special solvind tools avaliable:
• “Reduce” tool, that detects all free squares with only one possible value, fills them and repeats the process until no more of such squares are remaining.
• Limited “solve” tool, that tries to solve the sudoku using a backtracking algorithm. Unfortunately, the phone goes out of memory very soon, so this tool is very poor. I’m working to optimize this algorithm in order to enhace the decission path and to lower the memory footprint of the “context” of each step.
• Tested on Nokia 3100 and Sony Ericsson T630.

From now on, I would like to enhance the following points to reach version 1.2:

• Optimize the solving algorithm until it’s useful.
• Write a sudoku editor and give the user the ability to store input sudokus into RMS memory (resource management system) on the phone.
• Make possible for the user to push and pop the current game state on a stack, to do backtracking by hand.

And now, enjoy some screenshots. I’ll keep you informed. Stay tuned…

This night I’ve been doing some tests with Wireless Message API (JSR-120) that, surprisingly, is supported by my cheap Nokia 3100.

I was expecting that this API lets me examine the SMS inbox folder or something similar, but it doesn’t. It “only” provides a way of data interchange (as in a network socket) using SMSs as transport layer. You can open a connection to a url like “sms://666123456:5000” (yes, I don’t know what a port means :-m) and, on the other side, listen to “sms://:5000” and receive data.

I wrote a small test midlet and was able to “send” one of such messages (the phone showed me a confirmation dialog), but it never reach the destination phone. Maybe it can only be received by another Java midlet, I was expecting a normal SMS.

I’ll continue to do more testings when I’ve time enough…

P.S.: Dape has discovered today that WMA doesn’t make possible to access to sms inbox (for security reasons?), and can only be used to receive sms sent by another midlet using WMA. Phones that are able to receive WMA sms on their inbox are implementing the specification in a wrong way.