Older blog entries for movement (starting at number 67)

Mosfet has a rebuttal of Havoc Pennington's Free software UI article. It doesn't seem to be very well thought out, or even proof-read. For example Mosfet says it's important not to discount features and "snazzy graphics", but Havoc never once made any such suggestion - it is the over-reliance that is the problem. He purposefully mis-interprets Havoc's re-working of mpt's points and then attempts to prove that KDE hasn't fallen prey to them.

I'm not sure whether it was accidental that Mosfet mis-interpreted the bullet "Maintainers cave in and add lame preferences rather than endure flamewars.". His response to this is that "it always helps to listen to what the users say when implementing things". Who on earth isn't doing this ? The problem is when users are saying different things. What do you do ? Add a million preferences to satisfy every vocal user, or just stick with some sensible decision ? It's important to remember that the vocal user population is not a representative minority. This is the central point that Mosfet really does not get.

I don't think he understood what Havoc meant by "auto-generate" either. How exactly is Qt Designer automatic ? It's clearly not - it would be called Qt Automator otherwise ;) He doesn't seem to get the difference between configuration and preferences. Speaking of the latter, I wonder why he hasn't dealt with the point that the difficulty of a decent organisation of preferences is a factor of the number of preferences.

He has an unfortunate habit of directly contradicting himself too, which makes his points a little confused :

" Havoc states:

"The traditional, de facto free software line between when a preference should exist and when it shouldn't is "a preference should exist if someone implements it or asks for it." No one is going to seriously defend this one

I can and will! ... yes as a developer you do have to draw a line about what your application should or should not do "

Tried out Gnome 2.2 via Garnome today. It built fine, but, alas, completely failed to work. Starting nautilus from within KDE was so swap-happy (on a 512Mb machine) it took me five minutes to get to a shell to kill it. Xnest, exactly as described in the Garnome README, just gave a black screen. Starting gnome-session from a new X session just left the cpu pegged at 100% on the splash screen and didn't progress from there.

I don't know why Gnome hates me, but it does. Of course, every KDE release I've built has worked just fine.

What's the point in having QA on a bug tracker when QA don't even read the bug ?

The nutcase comments in that bug are pretty funny too.

Not sure why the histrionic crackpot mglazer rated me at 1; I've never complained about his rampant idiocy before. Perhaps it's because I whined about politics on recentlog...

Finally, irrefutable proof that any attempt to fix Mozilla's usability problems is pointless.

Thank the lord for Phoenix.

Was introduced to the phrase "joe-jobbing" by sarnold, and some spammers. Wondering about the etymology of that.

Random idea #1: the electric guitar is more essential to cheesy cinema than the camera is

Random idea #2: googlocracy: a state where every decision is decided solely by means of a googlefight.

Have I started dribbling yet ?

The rough x86isms of oprofile starting to show up on weird-ass platforms like PA-RISC. The fixes seem to be relatively simple luckily. Modules are a different matter. Rusty refuses to fix a regression in his new code where the original path of loaded modules is no longer available. This will no doubt lead to people insmodding ./blah.o and getting wrong profiles. All complaints shall be bounced to him (hey, he said to do so).

Been desperately trying to get Red Hat 8.0 to work properly with fonts. I was under the impression that using .UTF8 locales, stuff should just work. Alas, it's not the case :

• Cannot compose with en_GB.UTF8 in konsole at all (acknowledged in bugzilla)
• Cannot compose Latin L with accent in gnome-terminal
• Cyrillic_a is invisible in gnome-terminal, mis displayed in konsole (with LANG=ru_RU, it's completely wrong too)

and a host of more minor problems. It looks like this stuff wasn't very well tested, a pity. Still, it's a big step forward to ship with this stuff. Hopefully when it's not a .0 release, things will work :)

Argh, stop it with this "favelet" nonsense ! MPT is right - it's not only an unnecessary neologism when a term already existed, it's actually worse too. What the hell is "fave" about some javascript ?

Please, please - "bookmarklet" ! Thank you for your co-operation; feel free to return to your content.

moshez : actually, you raise an important point, and one that often leads to much confusion and misunderstanding. What does it mean to say that "vi is usable" ? Many of us find it indispensable to our working practices; next to bash, I would be lost without it. Yet, vi goes against all the received wisdom concerning usability. It is strongly modal; it does no presentation of available options; it has low levels of feedback, and its learning curve is ridiculously shallow.

The answer is of course obvious, but in two parts. First, there is the "zeroth law" of usability: know your target audience. vi is indeed completely unusable to non-geeks[1], but the target audience are people capable and willing to learn its interface. Second, a significant aspect of usability is the efficiency of the interface: vi is a blindingly, hugely efficient way of editing data. This is why we all love it so much.

Most people who accuse UI design advocates of wanting to "dumb down" interfaces start off biased from one of these two points. They either confuse efficiency with usability (one is part of the other, and not necessarily the most important part), or they assume that they are the intended audience. And cue massive flamage.

[1] sorry for the term, I know of none better

moshez : be serious ... lynx is essentially unusable for any non-geek. There are lots of reasons for this, in large part to do with graceless degradation of the mis-designed websites. Lord knows, Mozilla has its usability problems, but at least it stands a chance as a contender (and Phoenix^Wwhatever stands an even better chance).