Archive for February, 2009

Linux display drivers – we’re not yet there, but getting closer

Friday, February 27th, 2009

I had an Nvidia card all the time but lately (couple of years) I have found it pretty annoying with Linux: switching desktops became slow, widget redraw was sometimes visible, compiz backpuffering left some garbage on the screen, but the worst thing was that I had many segfaults with Amd64, some of which I really did trace back to the Nvidia binary blob.

Now I bought an ATI card and I experience some issues on Windows (most with the stupid Catalyst control center, which is just as annoying as the Nvidia counterpart and some more serious ones with monitor standby/suspend) so I thought I’ll give it a try with Linux (using the firegl binary driver). Is this any better than Nvidia? To my greatest dismay, it is not.

The speed is very good, opengl performance is great segfaults are gone; compositing is still not the best (but I guess this may not be a driver issue), xset dpms force standby works absolutely well, BUT using an opengl app on the desktop (Blender, anyone?) totally freaks out the desktop (does not depend on the window manager, broke apart Xfce, Gnome and Fluxbox), both with or without compiz. Finally using the dual head mode the font rendering gets messed up (fonts are extremely small), which seems to be a resolution/dpi problem, but I’m too lazy to look into it. Not to mention the fact that the Linux Catalyst Control Center is the ugliest app I have ever seen (okay, probably a motif application, blast from the past:)).

Website split testing – what they won’t tell you

Tuesday, February 17th, 2009

Some seo and marketing companies believe or state that the split testing (running two different versions of a page, pages or services on your website) or multivariate testing (certain page elements are being replaced and tested in paralell with other variants) is the magic bullet for increasing your rate of profit; but before we believe everything, let’s think twice: do people like malls and supermarkets? They do; Tesco, Salesbury or Auchan are popular examples of supermarkets where price is a major winning factor: one doesn’t go to Tesco because shopping is a “joy” – one goes to such place, because the quality is “pretty much okay” while the prices (and season sales) are unbeatable.

During split tests usability should be a major concern: an ad filled, blinking-scrolling website “probably” will not be as popular as a clear, well designed web shop – during these early periods the split test will show what we prety much know: I wouldn’t go to Tesco for a weekend shopping if it was a dirty, rotten place on the end of nowhere. Or I would: give me an unbeatable price and I might consider spending my money there (hence many butt ugly webshops with Joomla, Drupal and other home made CMSs exist) – with a better, comfortable environment the rate of return would clearly increase. But what happens if the services are fine, the layout is okay, navigation is user friendly – and you get all excited about testing small design elements here and there? Would I go to Tesco more often if they painted the doors red? Would I spend more money if they relabel all the products with shiny happy smily labels talking to me in a distinct voice, “have a good time here, dear customer”?

Probably I couldn’t care less. As long as the services are “okay” and I get what I want for my money, it’s just fine – treating the customers like a bunch of monkeys, trying out small “fine tunings” on your shop (let it be a webshop or a real store) will not boost those numbers; either because your monkeys don’t care (they already know where the banana is, how it looks like and the trees already look and feel pretty familiar) or just because monkeys and psychology don’t mix. You, as a designer, can tell yourself how your newly designed click-me button is so much better, or you, as a copy writer can tell yourself how your new copy makes the user want to click on the product, in the end you just convince yourself about what you want to hear. And what will happen in these cases with your numbers? There will be 1-2% fluctuations, nothing else. Have fun figuring out those numbers.

Patent trolls in the IT world

Sunday, February 15th, 2009

Beranger is ranting about patent trolls again (this time Kaspersky Labs) – so, while I find this whole patent trolling utterly stupid, I just gotta realize, I pretty much don’t care about it anymore. Linux is “mostly” protected as long as it is not mainstream and for the rest: give me a break, They won’t patent protect the freaking wheel (besides we’re talking about IT, bunch of ones and zeroes, it’s not human lives)! I’m not in the innovating business and if all hell breaks loose, like I don’t know, we end up having one OS only, well, it would suck. But that’s all. Many things suck yet we manage somehow, usually without fixing it, mostly because we’re just too tied up in our small crap powerstruggles from everyday life to corporation business.

Ubuntu X86_64 stability revisited (from time to time)

Friday, February 13th, 2009

Ubuntu x86_64 is indeed unbearably unstable (still, now Intrepid Ibex is the chosen one). It’s not that I give a damn, but it’s a tiny little bit annoying, how the 64 bit version on MY machine, for ME happens to end up having segfaults all the time. From firefox to gnome-settings-daemon, from opera to audacious, stuff I use – and it only goes away on a reboot. To make things clear: Arch 64 works, Debian 64 works, Windows-7 64 works (though not much fun), Windows 32 bit works. I know, I know, it works for everyone else, so shut up please.

Code quality and working in a team

Tuesday, February 3rd, 2009

I usually had been very harsh with others when I got to uncommented, undocumented, badly written code, especially in house – most of the time I thought about this as a matter of capability, but when I myself resort to low quality code I really have to stop for a moment and think twice before criticizing others. So far I have been trying hard not to argue with my boss (on matters that should not concern me or on matters I have no authority to comment on / change), but now this is the time for the same policy with colleagues.

Where specifications are most of the time scarce or badly written, where deployment is always the most important thing (even if the timesheet for the next week is still empty), where most people just don’t feel the need to comment the code (or just remove comments with refactoring), where insane ideas from management people pop up on a regular basis, when doing split tests on the live site for the sake of 1-2% lead increase is more important than thinking about how large scale parallel versions will affect the application logic itself, where search engine optimization comes way before code quality and user experience most of the time (pretty ironically) negating the means of standard, well-proven accessibility: I really wonder, how am I supposed to deliver high quality software? And what’s worse, noone really cares for the quality part (they would say “yes, we really do” wholeheartedly here, but caring for me rather means maintainability than rolling out new features as fast as possible). Without starting to figure out who’s right or wrong, which I really don’t feel like anymore, I really am close to understand someone who just wants to “get the job done” – while I myself am facing the fact that neither I, nor most of us have the qualities to produce “best quality” under such circumstances.

I do know how my boss would react and how impudent this all sounds (with keeping in mind that – fortunately – this is a for profit company, not a bunch a freetard hippies in a cave), but even though I’m doing my best, I felt very frustrating that no matter how hard I try, I cannot comply with my own standards, angry with myself and others, now I think it’s time to change and let things go on their own way. I still try not to break things, I still comment my files, I still try to write clean and readable code, but I feel less guilt with the terrible hacks I or other team members do day after day.