DeveloperWorkshopNovember2007Report

A summary of things we did, achieved, and tried at this years FAI Workshop - DeveloperWorkshopNovember2007

Introduction

In the end, quite some people who wanted to come did not arrive at the meeting because of several troubles. So, we were only 5 persons.

We discussed many things and got some work done.

Participant's summaries

Here a summary from every particpants view:

Christoffer Kugg

Personal goals

My long term goal with my contributions to FAI is to make it easier to understand use and implement. Therefore I am devoting a big part of my time spent on FAI to implementing icons graphics and writing information text to various interfaces in FAI.

My intentions for the work meeting was to mainly work with the FAI user interface. I wanted to get input and guidelines from other FAI developers. I wanted to have a graphical template for how it the interface should look like, done by the end of the weekend. Further I wanted to have it in some way accepted by the developer(s) whom will write the back end for the same interface. I wanted to have defined what should be needed to proceed the work on a automatic host detection tool for fai. I wanted to fix at least one bug.

My achievements

I managed to finish most of the templates for FAI User Interface and have got them accepted. During my upcoming evenings (after the event) I will polish them and write hint text to the interface to make it easier to understand.

I have a quite clear idea of how the client detection tool will look like and work and are now ready to start implementing parts of its functionality.

I fixed one minor bug in FAI related to licensing issues.

I also started working on a local FAI monitoring interface to be able to view a installation in a more civilised fashion than the current flashing text.

All in all I am very happy with the well spent time in Essen Linux Hotel and as usual I am very impressed by the work done by the rest of the FAI team.

Thank you FAI developers and contributors, and thanks to all sponsors whom made this happen.

Best of regards Christoffer Jerkeby

Thomas Lange

I had a list of small todo's which I mostly could finish or discuss with others. The discussion of several bugs in the BTS helped me a lot to know how to handle similar bugs in the future. I gave some hints to Michael about my coding standards for the new partition tool. I pointed Christoffer to some perl packages for rewriting my faimond that run on the local machine completely in perl (currently using ugly systme calls).

We all got good introduction to OO programming in Perl and about the Corebiz software.

There are several topic we did not work on like I expected (e.g. roadmap for lenny, wiki review, new config hierachy, squashfs for fai-cd).

I was impressed about the implementation of simple tests via hooks and the progress for StorageMagic which will be included in 3.3.

Summary: a great meeting. We should think about doing this more than once a year, even if not a lot of people can attend.

Michael Tautschnig

For the workshop it hoped to get the new partitioning tool ready for integration with mainline FAI. However, as software development often fails to meet the timelines, this was not fully achieved. Instead, the following advances happend:

• Mrfai had already reviewed the entire code and suggested a number of changes, primarily of cosmetic nature. During the workshop and in the week following it, these were mostly integrated, apart from splitting the larger functions into smaller pieces.
• The pretty large data structures used inside the perl scripts would call for an object oriented approach. After Ingo Wichmann's short introduction to OO perl coding, lazyb0y, michael and Ingo had an extensive discussion about viable designs following this paradigm. However, no satisfying solution has been found yet and it was agreed that the current (non-OO) design should serve as the first implementation. Later, the tool will slowly migrate to an object-oriented code base. This decision was mainly based on the idea that the user interface would not be change in such a (partial) re-implementation.
• To improve the quality of the tool and FAI in general, Sebastian proposed a regression testing suite. This is essentially a config space that has dedicated testing classes an scripts to perform checks on the installed system.
• Finally, I was decided to give the tool a new name, now calling it Storage Magic