Announcing the availability of SetiContainer. The SetiContainer class is a C++ class that allows convenient access to all the information in the state files of a SETI@home client. It periodically examines these files and updates its own data members accordingly. Other programs can use this class to extract information about the progress of the client, the found signals, or the work unit being processed. Additional functions offer more advanced info (like the amount of TeraFlops in a work unit, estimated time of completion, etc). The SetiContainer class is based on the Qt toolkit, and extensively uses Qt's Signal/Slot technique.

This class comes with more or less detailed documentation. Even if you are not interested in coding, you may find some information about the state files of the SETI@home client. Visit the Download page to get the SetiContainer package. Opinions, suggestions, and additions are welcome.

To all of you, Merry Christmas and Happy Holidays.

There is one point I forgot to mention yesterday. Some users have already noticed it: SourceForge membership is not required any more to use the public forums. You can now post messages as "Nobody". In any case have a look, because the forums have turned into a valuable resource of information.

Activity has been quite low on this page recently, so I think it's time for a brief status report. Ksetiwatch celebrated its first anniversary a few weeks ago, and I decided it's a good idea to clean up all the mess in the code that accumulated during the past year. A nice side-effect of this work is a new C++ class, the SetiContainer class, which allows convenient access to all data in the S@h client output files. I'm planning to release this class as a separate module, so that other projects can benefit from it (is anybody willing to write an applet for KDE's panel?). Unfortunately, SourceForge is in the process of moving their servers to a new location, so that file releases aren't possible at the moment. Well, this gives me more time to work on the documentation.

Speaking of SourceForge, because of the aforementioned relocation of the servers, this site will not be available on Friday, December 15 from 9pm PST until 3am PST.

Back to Ksetiwatch: two users noticed that all recent versions of Ksetiwatch seem to eat memory. Not very much, but when letting it run continuously for two weeks, the total memory consumption could sum up to over 30 MBytes. After some digging in the code, I could find the reason for this memleak. A fixed (but still unstable) version is available in the CVS repository. Those who don't want to update from CVS I recommend to restart Ksetiwatch from time to time. Many thanks to Roeland Th. Jansen and Jan Knutar for pointing out this bug. I'd like to release a bugfix version, but... read above.

After a quite long period of silence, I'm back with a new version. The most important addition is the graphical display of pulses and triplets. Actually, I had this finished more than two weeks ago, but was set back by some severe hardware problems causing my system to spontaneously reboot at frequent intervals (it turned out to be a faulty power supply, BTW). Anyway, here is a list with the most important changes:

• Pulse and triplet graphs. I rearranged the popup menu in the Analysis tab a bit to accomodate the new feature.

• Just like for peaks and gaussians, you will now get a notification when new pulses or triplets are found.

• The fit curve in the gaussian graph is drawn much smoother now. You will notice this for narrow gaussians (having a low sigma value).

• When the processing of a WU is restarted (e.g. because of upgrading to a new S@h client), the max scores for peak, gaussian, pulses, and triplets are reset to zero.

• Graph widgets remain hidden when they are closed. This reduces memory footprint (previously, widgets were discarded without being deleted).

• Since for triplets 'power' and 'score' are obviously identical, I made the triplet entries in all lists shorter. Only 'score' and 'period' are shown.

• There was a bug in the summary list resulting in garbage output for triplet data; fixed.

• Converting old SETILog.csv files didn't work properly. Instead of writing a SETILog.old file in the S@h folder, something like 'setiathomeSETILog.old' was written to the user's home directory. I added a missing / to the backup file string.