Posts 

With the help of other pim developers, I updated the kdepim tasks page. The kdepim tasks page was a very useful tool during the last release cycle, and is already helping the kdepim quality team and eventual newcomers to quickly see what needs to be done. Yes, we have a strong kdepim quality team, I am glad to say it. There are several people working on different areas of the pim applications, including the docs, websites, etc... who work silently, and are not explicitely involved with the quality team. However, at leat two guys, (me and Antonio Salazar), are already jumping in when needed. Antonio worked on Kontact docs during the last release cycle, and is working with KOrganizer whatsthis and hopefully with the docs too, and I wrote KPilot docs and whats this back then, and will do whatever is needed for this cycle (I did not decide yet). Some others started helping out in the last release cycle and submitted smaller contributions, and I hope to hear from them again (Hi Ramon and Jörg, how are you doing?). This effort started with the quality team annoncement six months ago. Unfortunately, it seems that other modules did not have the same results as we did. My email to the quality list about the announcement for this release cycle got no response. I offered help: "If you want to create or update your page to the next release cycle, I can help! Just drop me a line.". Nobody asked (yet). So if you think the tasks pages are a good idea, and you want to participate in the new release tasks annoncement, please drop me a line, or if you don't, I will go with with kdepim only. Read More

Two of my favourite books about computer programmers are 'The Psychology of Computer Programming' and 'Understanding the Professional Programmer' by Gerald Weinberg. So I was interested to read this short interview with him. Read More

first of all I hope you are aware that your heading "a lesson to all" sounds like pure arrogance to everyone who took part in these discussions but is still not considered in any of your reasonings. Read More

I recently had this email exchange with my friend Geoff. I don't need to add any further commentary, but I've personally bet the farm on custom application development. Hmm.. He's what we said anyway: Read More

From the One-Thing-Leads-To-Another Department: You may recall from a previous blog that I am changing libkcal to provide a better Incidence sorting interface. While I am working this I decided to check for consistency across the Event, Todo, and Journal methods in libkcal. So, I searched for an on-line KDEPIM API Reference at the KDE API Reference pages. Not there. Read More

In the last two weeks Fabian has made huge progress with FreeNX: he designed and implemented a new security model for FreeNX (with the help of some outstanding people who are now regularly joining debates in the #nx IRC channel on freenode). It doesn't use the nxssh binary any more (which made some Linux distro security auditors to be very suspicious), but instead uses the most recent "standard" OpenSSH package. So any newly discovered future SSH vulnerability doesnt need to be an "extra" NX concern -- fix SSH and you also fixed NX. We hope to have made the security experts (notably those from SUSE) happy with this and that the upcoming audits of FreeNX will soon be passed without need for much re-design. he made FreeNX behave well with various types of clients (KDE's initial knx client for NX and FreeNX sessions, as available from KDE-CVS in the kdenonbeta module -- compile instructions are here; the NoMachine commercial -- but free-as-in-beer -- NX clients of the 1.3.x as well as the various 1.4-snapshot releases). For the NoMachine 1.3.x NX Client he even succeeded to hack an "auto-resume" feature into the FreeNX server: the user restores any suspended session automatically upon re-connection (the 1.3.x clients dont normally support that feature, they always create a new session). he made FreeNX use server-specific SSH key authentication (for the special "nx" user who initiates each connection) as well as the general NoMachine key authentication (for the "nx" user), as well as supporting a passwordless, key-based SSH connection without the need for the "nx" user), as well as PAM-based authentication schemes. But the coolest thing last.... It was inspired by a Slashdot posting (yeah, sometimes you even find gems there!) from someone who wished he could use an "ssh -NX" instead of an "ssh -X" commandline. This is not exactly here yet. But what you *can* use now is one of these: nxtunnel username@remote.NX.host bash nxtunnel username@remote.NX.host xterm This will start a Bash shell or an xterm (just as if you used "ssh -X username@remote.NX.host xterm". But here the session is tunneled through an NX link, going through SSH. If you now type into the xterm or behind the new bash shell prompt something like "konqueror" or "kmail", you'll start these applications on the remote end (and display them locally), speed-boosted by NX compression (which is better and less CPU-intensive than generic ZLIB compression) and by NX caching (which is unprecedented in its efficiency and rate of cache hits). This is already considerably faster than "ssh -X -C" sessions. The third NX component, the X roundtrip suppression, which makes things really fly, is also enabled in nxtunnel sessions... [CAVEAT! Be aware that the mentioned third element of NX's superior methods to speed up remote X GUI sessions, its round-trip suppression scheme, does not yet work very reliably for the case of single application windows. The reason is, that the "nxagent" part of the NoMachine software currently only has experimental support for "rootless windows". Single applications therefore normally bypass the "nxagent" part of the NX machinerie and still suffer from the roundtrips. (If you are interested in a flowchart of the NX architecture, look at this one While benefitting from NX compression and caching, a roundtrip-bogged NX session is still faster than a plain "ssh -X -C" session, but it is far from being regarded as as a ground-breaking thing. -- The nxtunnel script enables the rootless window mode with roundtrip suppression, but it has still some bugs. The most important one being, that if you close just one window (like any child dialog) using the window manager "close" button, it'll close the whole nxagent connection. But this is being worked on... To partially avoid this, exit or kill the application from the bash or xterm window you started it from.] Read More

I'm writing this on the flight from Tokyo to Sydney, with intent to publish later. Earlier today I flew the Frankfurt to Tokyo legs, so anything in this blog needs to be understood in the context of some potential jet-lag :-( Read More

Many software developers stumble over the "Joel On Software" columns at some time and like them. Sure, Joel was a Microsoft employee and he develops proprietary software, but still his columns are inspiring and fun to read. Sometimes he is completely wrong, though. Read More

Brad is commenting about a research project that does "Internet Suspend / Resume" of user sessions: "...interesting idea - they basically envisage a thick-client model, running on a virtual machine (VMWare in their tests) which can suspend. The suspend state then gets shipped over a network connection to another machine where it can be “resumed

So, I've been working on some kdevelop language support C# for the past few days. I already have some pretty interesting screenshots, but I recently discovered that KDevelop is going through a major change in the way it handles different build systems. Roberto Raggi is working on a New Project Manager (NPM) that will form the basis of a common project editor widget for all the various languages. Individual language/buildsystems will then customize the interfaces to this NPM in order to provide language/buildsystem specific features. It is a great idea as it will cut down on the duplication of code in KDevelop and provide some standard and uniform UI widgets. Since I found out about it, I've been porting my initial C# build part to this new architecture. More... Read More