KDE Blogs 

One thing that recent dicussions about usability have highlighted is how much our decisions are currently based on personal opinions rather than hard facts. Aaron has recently been trying to address this by asking people how much they use the various buttons in the konqueror toolbar, but this is both labour intensive and inaccurate. So, I've written the 'Usage Monitor' plugin. The Usage Monitor is a KParts plugin that records a log message whenever a KAction is fired. The log simply notes the action that was activated and how it was done - eg. was it via the toolbar, a menu or a keyboard shortcut. By analysing this log it is possible to measure which parts of an applications GUI are being used, and which are not. If a reasonable users can provide sample data using this tool, then we can build up an acurate picture of things we could previously on guess at. The plugin itself is tiny at around 150 lines of code, and works with any application that supports KPart plugins (which covers most KDE apps these days). No modifications are needed to the application itself, and you don't need to tweak the monitor for your app either. The code is a good example of the power of the XMLGUI framework as it shows how easy it makes creating tools that can apply throughout KDE. In this case, it even lets you track the use of actions provided by plugins and embeded kparts as well as the main application. Here is a sample log recorded from Konqueror: Read More

Well now that I have been working with getting Kolab to handle my spam and virus issues I think I have finally found a nice solution. A few weeks ago when I was testing my email spam filters here I went out on usenet and posted my name a few times. Now that I get about 1500 spam messages a day I have a very nice test set ;) Read More

A common complaint I get from developers is that their work isn't showing up in the digest. I'll explain how the whole thing is done, and then get to the question. I go through the 2000 or so commit emails from kde-cvs list, and select commits that are either bugfixes, new features, optimizations or security fixes. With bugfixes, any that have a bug number are selected. I then run a script that builds the html, statistics and other stuff, edit, add a few things here and there, then publish. This works reasonably well, although there are imo too many trivial bugfixes highlighted. Read More

Aaron has posted a blog entry about developes 'fearing' usability teams, I've replied to his post itself by a comment but I'd like to provide a fuller view of things here. Read More

ok... so as the recent thread on kde-core-devel re: kcfg showed, people FEAR usability efforts. that's right, people FEAR it. that word was actually used in relation to the topic. they are afraid some people will screw it up, that KDE can't cope with a drive towards being more usable without becoming pure shit. i really feel like ranting about thickheads, but that would just be adrenaline speaking. as much as that would feel good ;-) , i'm going to try to be constructive instead: what can be done to gain the trust of people in KDE? or is that a pipedream? Read More

ah... finally in my new house and it's feeling good... now to start putting the wraps on various bits of code i have laying around ... committed small things to kjots and kcm_kicker the last two days, and have bigger chunks of kscd, kicker, TOM and systray patches awaiting my attentions. i also have that bookmark code lypie's been patiently waiting for me to commit. =) Read More

ok.. so, i'm sick of waiting several seconds for kwrite to pop up whenever i do a View Source in konqueror. yes, one might be correct in saying kwrite should open faster. but i've got a different idea: why not have a simple viewer window that runs IN the konqi process ala Mozilla? i've had this on my slower box at home for some time now and BOY does it make a nice difference speed-wise. i need to add some basic viewer functionality to it like Search and Print. perhaps just a nice little toolbar with a couple of buttons on it (and an incremental search line? hrm...) anyways... i'm pondering whether it's worth polishing up and attempting to convince the Konqi maintainers that this is a better approach than launching an external editor. here's my benefit/cost summary: Read More

Being on holidays certainly gives you time to reflect on what new features you'd like in your text editor :) I've come up with something which I hope will help make using a text editor a better experience by providing more assistance to the user... input filtering. It came to me when I was thinking of cleaner ways to do code templates. At the moment, I think you have to use a keyboard shortcut to insert code templates? Anyway, this will make it easier for the 3rd party app developer to e.g. respond to "for " and insert a template. It will also make immutable text a possibility, something requested in a comment to my last blog. In fact it will make it possible to enforce valid structure on a document, though I personally think that would be too restrictive. Immutable text could be appropriately highlighted, and the cursor could move around it in a logical manner, so that the for ( ; ; ) { } template could be navigated more easily. Other ideas on the go are improved text navigation eg. easier use of code folding (previews of folded code? auto-expand modes?) and the ability to add a custom widget to the top and/or bottom right-hand corners. I've also decided to improve the idea I had for syntax highlighting by giving each text range a "state" variable, and associate a specific highlighting style to a state, then the developer just has to change the state of a range to make it look different. I'm also taking requests for whatever text decorations you'd like to see available in the next kate - I'm thinking of borrowing some ideas from css, just let me know which ones you think would be useful. Furthermore, I'm planning to extend on code completion by adding syntax highlighting to the dropdown, expanding the info presented, allowing several different filtering and sorting rules (eg. by inheritance; by access (public/protected etc); alphanumerical; and hiding / disabling completions which don't fit the current context), and integrating (somehow - ideas?) summary documentation for the highlighted method. My flight calls, gotta go... someone should sponsor KDE developers going on long-haul international flights so they can hack from business class! Economy is way too cramped...

My ISP confirmed last night that my ADSL should be working from tomorrow, so I'll be getting back to work on XPath and KJSEmbed. Hurray!

I've committed Qt bindings for libusb - see http://sourceforge.net/projects/libusb/ This includes API docs and unit tests. This isn't released yet, so you'll need to suck down the CVS version. I'm a bit worried that the bindings aren't really Qt-ish enough, because I'm not very experienced with class design. So I'm taking advice (and patches!). You'll need to provide fairly specific advice though - at least the interface needs to be specific. Read More