Interesting Python speedup idea seen on planet.gnome.org.

Unfortunately using it I found 2 things:
1) all existing python scripts that do /usr/bin/python or /usr/bin/env python have to be modified (?)
2) currently it seems broken for any python app that does local imports, meaning it can't find 'findrox'!!

I may try to fix these and report later...but I wonder if this concept could be added to the Panel somehow.

I've made the first release of 0publish-gui today. This is a GTK utility for creating and editing Zero Install feeds. This should make it a lot easier to get started. Here's a tutorial with lots of screenshots. Have fun!

I've been doing some work on the GUI for 0compile, used to build programs from source. Here are some screen-shots showing how we can download and compile ROX-CLib.

I've been experimenting with a way to combine FUSE/SSHFS with ROX AppDirs and its support of mountpoints.

I have a hackish version now that works well enough for me. Perhaps others could suggest improvements.

Prerequisites:

ROX Filer, FUSE, SSHFS, and password-less ssh access to the remote host (see ssh-keygen and .ssh/authorized_keys).

The last part is required to keep the Filer/Desktop from hanging while waiting for a password to be entered.

Unmounting:

The standard umount command will not unmount fuse-mounted mountpoints. You need to use fusermount -u for that (or be root?). So, using the new configurable mount/umount commands in ROX-Filer 2.5, I created the following myumount command and put it in my path and set ROX-Filer to use this in place of umount.

Three weeks ago, after the large amount of comment-spam we were getting, I set it up to record spam attempts in the database. I've created a new known spammers page with the results.

For individual IP addresses, TELEFONICA DE ESPANA is currently winning with 1085 attempts, easily beating Microsoft into 5th place (just 151 attempts). I suspect the picture changes if we group by organisation though, since many of them are using dynamic IP addresses. Any PERL hackers want to write a script to run the results through whois so we can see a proper breakdown?

Watching how other people use programs often shows you tips and tricks you didn't know. ROX-Filer has lots of features; some are extremely useful while some I almost never use. Here are my favourite features and settings...

I was looking for an excuse to play with the new Google hosting service. So I put WiFi there - the reasoning being that it wasn't my code to begin with and if someone else wants to contribute it is a bit easier to give others access (than on my personal site).

I wonder if we should put other abandoned ROX stuff there.

In case you haven't checked it out (rox-wifi) they're offering a simple Home Page, Keywords, SVN access (with very simple web view), and an Issue Tracker with a very simple interface. Discussions can be linked to Google Groups.

Prolific GNOME profiler Federico has been profiling GNOME usability. The results? Saving is a big usability problem. The list of problems sounds almost identical to the motivation for the ROX saving system, itself based on RISC OS. RISC OS users never save files without knowing where, because they have to drag them where they want them.

Imagine a new office worker who tries to file an important document by dropping it on the floor. Pretty quickly, someone will yell at them to learn the office filing system. ROX does the same. Here's what happens if you just enter a name for a new file and press Return in ROX:

Stumbling on Jim Propp's self-referential aptitude test reminded me that I should really get around to learning Alloy (a "simple structural modeling language based on first-order logic", according to its homepage). The examples in the Alloy tutorial work well enough, but I wanted to know whether will also solve problems not chosen by the author ;-)

My final solution came out surprisingly neatly for a first attempt, following the format of the quiz closely, although it took much experimenting to get the syntax exactly right. The only bit I'm unhappy about is the helper predicates for question 12, as I couldn't find an easy way to test for primes, etc. Any Alloy experts out there want to show me a quicker way?

Wondering why X takes so long to start up (and inspired by the smaller source tree since it went modular ;-), I did a little profiling tonight. Turns out that probing the PCI bus is taking quite a while (Debian/unstable's X.org). The last call to pciFindNext() was taking 1.7 seconds just to report no more devices!