akademy is done. wow, what a week!
yesterday we had the last set of lightening talks and again some very cool technologies were shown. it looks like the dream of having a usable, powerful and beautiful universal viewer for kde4 is a reality with okular. it isn't the only "universal viewer" (which is different than a "universal component embedder") out there, but it is the most complete and beautiful one i've seen. it's also pretty damn fast. there are some ui bugs still apparent in it, so thankfully there are some months before kde4 ;)
the global roaming, organs identity ui system and gamefu (think amarok for console emulators) were also fun ...
today those who are left (many are already on their way home) are sitting around mostly quietly hacking. this is interesting because much of the past week was crammed full of BoF's, meetings and what not... we had a lot of communication to do and that cut into the hacking time. as kde takes on bigger problems with more finesse and expands into taking care of things we've neglected more than we should have such as our websites we have more need to coordinate.
to catch up this slack i'll be working with knut, the new-ish trolltech community manager, and others in the community to pull together 5-7 developer meetings over the next year. each will be focussed on a topic and aim to build on the coordination of akademy with code (or whatever the deliverable of the group is).
in any case, right now i feel like crap. i caught someone's flu and feel slightly eviscerated right now. flying home tomorrow will be fun(tm).
Saturday, September 30, 2006
Thursday, September 28, 2006
lightening talks
there are three lightening talk sessions scheduled for this week: one during the odf day, one today and one tomorrow. i chaired the one today and it was about kde in general. unfortunately they weren't taped (only the contrib conference was).
the presenters were:
Torsten Rahn on Marble, a 3D geography planner
Stephan Binner on Kerry, a KDE frontend to beagle
Thomas Zander on Flake, the new document component model in koffice2
Alfredo Beaumont on KFormula, the amazingly complete mathml component he wrote as part of a summer of code project
Kevin Krammer on QtCenter.org, a community forum for people writing software with qt
Nadeem Hassan on his kde4 SIP library
Junkyu Park on KLDraw, a 3D opengl lego designer
Jos van den Oever on Strigi, his really cool indexing engine (he also did a BoF on it earlier in the week)
i was particularly impressed with the mathml app which renders some 70-80% of mathml properly and supporters even more than that with various rendering glitches, whereas the open office mathml tool only does some 20-30% of the mathml standard.
Junkyu's (segfault on irc) app was similarly impressive. he claimed it to be "unfinished" and then loaded up a very complex lego model with hundreds of pieces in three dimensions, moved pieces around, built sub models interactively and then rendered it to a high quality via povray. he has plans for blender as well and already supports the full ldraw standard in all 2500 pieces glory. amazing. he promises to release it soon for all of us lego addicts to play with. it's also the only ldraw app for linux/unix.
looking forward to tomorrow's lightening session.
the presenters were:
Torsten Rahn on Marble, a 3D geography planner
Stephan Binner on Kerry, a KDE frontend to beagle
Thomas Zander on Flake, the new document component model in koffice2
Alfredo Beaumont on KFormula, the amazingly complete mathml component he wrote as part of a summer of code project
Kevin Krammer on QtCenter.org, a community forum for people writing software with qt
Nadeem Hassan on his kde4 SIP library
Junkyu Park on KLDraw, a 3D opengl lego designer
Jos van den Oever on Strigi, his really cool indexing engine (he also did a BoF on it earlier in the week)
i was particularly impressed with the mathml app which renders some 70-80% of mathml properly and supporters even more than that with various rendering glitches, whereas the open office mathml tool only does some 20-30% of the mathml standard.
Junkyu's (segfault on irc) app was similarly impressive. he claimed it to be "unfinished" and then loaded up a very complex lego model with hundreds of pieces in three dimensions, moved pieces around, built sub models interactively and then rendered it to a high quality via povray. he has plans for blender as well and already supports the full ldraw standard in all 2500 pieces glory. amazing. he promises to release it soon for all of us lego addicts to play with. it's also the only ldraw app for linux/unix.
looking forward to tomorrow's lightening session.
kde in asia
the kde asia track was amazing this week. i had the pleasure and honour to meet with people from korea, cambodia, china and india who are working with and on kde. this included our kde india spearhead (who gave me a cool kde india t-shirt... thanks pradeepto!), korean community developers working on 3d applications and people from red flag and asianux.
as an interesting example red flag linux employs 30+ desktop software developers. they have a huge number of patches that have yet to be pushed upstream and it's a real investment/liability issue for them. so we'll be working much closer with them on syncing development where it makes sense. given their large deployment numbers, the potential market in their region and the amount of investment they are making in development, this is a very exciting opportunity.
one of the other outcomes is the new kde-asia mailing list, the start of planning for an asian kde event and some additional communications and marketing support such as a boothbox for asian events and sponsorship outreach.
this peek into the efforts happening in asia was truly eye opening. helped me realize that world domination is one step closer than i was aware of. =)
as an interesting example red flag linux employs 30+ desktop software developers. they have a huge number of patches that have yet to be pushed upstream and it's a real investment/liability issue for them. so we'll be working much closer with them on syncing development where it makes sense. given their large deployment numbers, the potential market in their region and the amount of investment they are making in development, this is a very exciting opportunity.
one of the other outcomes is the new kde-asia mailing list, the start of planning for an asian kde event and some additional communications and marketing support such as a boothbox for asian events and sponsorship outreach.
this peek into the efforts happening in asia was truly eye opening. helped me realize that world domination is one step closer than i was aware of. =)
hawaii in dublin
the other night we went out on the town a bit after dinner. not knowing where to go we did what we usually do and asked the nearest bunch of young women who looked like they were heading out for a night of fun themselves. they tend to know the Good Places(tm). in this case they were dressed up in faux hawaiian gear: an obvious tip off! they were heading to a hawaiian night party at a local club, informed us it was going to be great and invited us to join them.
and so we wound through the streets of dublin in the dark following these three laughing and generally wound up girls to the destination only to arrive and be asked by the bouncer for our student id cards. apparently it was assumed that we were students by our tour guides. unfortunately i'm not a student these days and so had to retreat back down the street.
however, there was a continuous parade of people in hawaiian gear marching down the street and virtually none of them knew where they were actually supposed to end up at. so i became an impromptu traffic cop directing people down the proper side streets. i received a few more invites to join them, including a couple of girls who insisted that the bouncer wouldn't remember my face and would let me in if i was with them. i didn't feel like ditching my fellow kde'rs nor did i feel like getting bounced at the door a second time in the same night and had to decline despite their persistence. tom chance was nice enough to offer me his student id for the night but that seemed a bit cheesy, not to mention i'm not sure i'd get by with his photo on it. =P
i'm back in saddle today at 09:00 working on various things. today is blocked out for some work with our communications team and the rest is for development on icon management. screenshots tonight perhaps.
tons of hacking is happening in the coding rooms, but due to my wireless being more dependable than my wired networking and there not being wifi in the coding rooms i'm relegated to the quieter conference rooms. however, quiet is good.
... and if you haven't caught it one of the other blogs yet, kde made an appearance on the t.v. show "heroes" the other night. neat.
and so we wound through the streets of dublin in the dark following these three laughing and generally wound up girls to the destination only to arrive and be asked by the bouncer for our student id cards. apparently it was assumed that we were students by our tour guides. unfortunately i'm not a student these days and so had to retreat back down the street.
however, there was a continuous parade of people in hawaiian gear marching down the street and virtually none of them knew where they were actually supposed to end up at. so i became an impromptu traffic cop directing people down the proper side streets. i received a few more invites to join them, including a couple of girls who insisted that the bouncer wouldn't remember my face and would let me in if i was with them. i didn't feel like ditching my fellow kde'rs nor did i feel like getting bounced at the door a second time in the same night and had to decline despite their persistence. tom chance was nice enough to offer me his student id for the night but that seemed a bit cheesy, not to mention i'm not sure i'd get by with his photo on it. =P
i'm back in saddle today at 09:00 working on various things. today is blocked out for some work with our communications team and the rest is for development on icon management. screenshots tonight perhaps.
tons of hacking is happening in the coding rooms, but due to my wireless being more dependable than my wired networking and there not being wifi in the coding rooms i'm relegated to the quieter conference rooms. however, quiet is good.
... and if you haven't caught it one of the other blogs yet, kde made an appearance on the t.v. show "heroes" the other night. neat.
Monday, September 25, 2006
nothing like a day of meetings
yesterday was the end of the contributor conference. the akademy awards were also given out yesterday. the feeling in the air was electric as i wrapped up the conference with an improvised closing speech. the applause for those who made akademy possible, such as marcus furlong and his team, was sustained and loud. it was a fitting end to two brilliant days of solid technical presentations.
today has been spent entirely in the annual general meeting of the kde e.v. it's like a booster shot of administration. it leaves a bit of an itch, but it's for the best. ;)
the best news is that tonight we have a function put on by google to take the string out and then it's on to the really exciting stuff for the rest of the week: the hackfest. well, that's the exciting stuff for me anyways =)

custom house near trinity uni where we're at
today has been spent entirely in the annual general meeting of the kde e.v. it's like a booster shot of administration. it leaves a bit of an itch, but it's for the best. ;)
the best news is that tonight we have a function put on by google to take the string out and then it's on to the really exciting stuff for the rest of the week: the hackfest. well, that's the exciting stuff for me anyways =)

custom house near trinity uni where we're at
Saturday, September 23, 2006
some quick thanks
some huge thank-you's and "you ROCK!"s (note the use of capitals? that's how much i mean it ;) to those who have helped make this whole akademy 06 thing come off with remarkably few hitches and a ton of energy:
tink bastian who really did a great job helping lead efforts for akademy coordination. her work with the sponsors was great, and she's been graceful enough to let us steal her husband for another week while he attends.
marcus furlong, for beingsilly brave and brilliant enough to take on hosting akademy. his support group has also shown themselves to be a crack team. there's even some people from brazil on it!
cornelius schumacher for doing a ton of work with the kde akademy team and triaging efforts between the e.v. board and them.
and of course all the sponsors and other volunteers, who have helped make this happen.
btw, the marketing team looks like it's going to flex its muscle nicely this week as well. watchout!
tink bastian who really did a great job helping lead efforts for akademy coordination. her work with the sponsors was great, and she's been graceful enough to let us steal her husband for another week while he attends.
marcus furlong, for being
cornelius schumacher for doing a ton of work with the kde akademy team and triaging efforts between the e.v. board and them.
and of course all the sponsors and other volunteers, who have helped make this happen.
btw, the marketing team looks like it's going to flex its muscle nicely this week as well. watchout!
akademy '06 day 0
so far this akademy is the best i've been to. why? the energy is terrific, the people are all quite active and the presentations have been kick ass.

a few of the attendees
we've seen examples of gapless playing of video in 10 lines of code or so, how voip stacks are coming together, hardware information handling, advanced desktop groupware data management, usability engineering process and progress, khtml status ... wow. i haven't seen this much progress in kde in a while. it's like sweeping the dirt off the remains of a bonfire to find it still blazing along.
the frameworks that are coming along are pretty mindblowing. just as kparts brought us things like 10 lines of code to a web browser, these frameworks are opening up the same kind of flexibility and power for other desktop services: media, hardware, desktop widgets and control, voip, messaging, calendaring .....
khtml is going to be killer post-unforking. between safari, nokie, omni, konqueror it looks possible to hit 10% market share for khtml. the feature wins for plugins, svg and more are extremely tantalizing as well. lots of work left on the "unmergification" but it's progressing very well (it renders pages using Qt4 nicely) and generally looks very promising both technically and strategically.
something like kde4 takes a while to get going, but it's finally starting to emerge. that said, there are a ton of items to be taken care of. i was struck during the talks as everyone listed their to-do lists and noted that they could use more developer help to make things progress faster that we ought to collate all these items on the web and start pointing people to it. something to coordinate this evening and tomorrow methinks.
the asia track was also very enlightening. we've got a great foothold in cambodia with the kde localization in that country going amazingly well as well as in india with our 10 month old regional group there. having developers from korea, india, china and cambodia here is helping us all understand the cultural issues as well as the technical. some of the work they are doing is amazing; there are terrific pools of talent and energy that we are only starting to discover when it comes to free software in these areas of the world.
and of course, one of my favourite things about being at a kde event is the massive icecream cluster that follows:

you scream, i scream, we all scream for icecream!
i had it to myself during some of the sessions today as everyone else was being good and paying attention ;) i configured, compiled and installed all of kdebase from scratch in somewhere between 10 and 15 minutes. nice.
i should probably blog about my keynote earlier today. but i'll save that for later. =)

a few of the attendees
we've seen examples of gapless playing of video in 10 lines of code or so, how voip stacks are coming together, hardware information handling, advanced desktop groupware data management, usability engineering process and progress, khtml status ... wow. i haven't seen this much progress in kde in a while. it's like sweeping the dirt off the remains of a bonfire to find it still blazing along.
the frameworks that are coming along are pretty mindblowing. just as kparts brought us things like 10 lines of code to a web browser, these frameworks are opening up the same kind of flexibility and power for other desktop services: media, hardware, desktop widgets and control, voip, messaging, calendaring .....
khtml is going to be killer post-unforking. between safari, nokie, omni, konqueror it looks possible to hit 10% market share for khtml. the feature wins for plugins, svg and more are extremely tantalizing as well. lots of work left on the "unmergification" but it's progressing very well (it renders pages using Qt4 nicely) and generally looks very promising both technically and strategically.
something like kde4 takes a while to get going, but it's finally starting to emerge. that said, there are a ton of items to be taken care of. i was struck during the talks as everyone listed their to-do lists and noted that they could use more developer help to make things progress faster that we ought to collate all these items on the web and start pointing people to it. something to coordinate this evening and tomorrow methinks.
the asia track was also very enlightening. we've got a great foothold in cambodia with the kde localization in that country going amazingly well as well as in india with our 10 month old regional group there. having developers from korea, india, china and cambodia here is helping us all understand the cultural issues as well as the technical. some of the work they are doing is amazing; there are terrific pools of talent and energy that we are only starting to discover when it comes to free software in these areas of the world.
and of course, one of my favourite things about being at a kde event is the massive icecream cluster that follows:

you scream, i scream, we all scream for icecream!
i had it to myself during some of the sessions today as everyone else was being good and paying attention ;) i configured, compiled and installed all of kdebase from scratch in somewhere between 10 and 15 minutes. nice.
i should probably blog about my keynote earlier today. but i'll save that for later. =)
Wednesday, September 13, 2006
it's like that "smack the gophers" game
spent time yesterday plugging a few more problems in kde3's kicker. both were my own making so i deserved it ;) one issue was positioning of centered panels when there was a panel on an adjacent side of the screen, the other had to do with the system tray spasming due to the layout resizing the tray in unexpected ways when it was put into certain configurations, such as a 49px wide vertical panel. i moved to a qlayout for the tray in 3.5 specifically to remedy other geometry problems but it introduced some new subtle issues. the internal complexity of kicker is a real pain. and watching it call resizing and layout function for each and every applet 4, 5, 6 or more times in a row just makes me cringe. oh well ...
i also realized yesterday that i hadn't received my tickets for my flight to akademy. so i went to the website i bought them on (travelocity, if anyone cares) and the itinerary was there and said it was all good. so i phoned to find out what had happened, assuming the tickets simply had been misrouted. turns out my reservation had been cancelled. lovely. two hours later i have a new reservation that's nearly the same price i booked for originally on a better carrier. it wasn't easy to beat it into their callcenter staff that selling me a ticket they couldn't provide and then not updating the itinerary on the website was something they should take some responsibility for. *sigh*
and then i upgraded the stable kde on the laptop from the kubuntu repos and wow ... lots of little regressions. amarok crapped out a few times before finding its feet and boy are those new icons not so great. but more on that in a bit. i had problems with fonts, kontact and more after the upgrade ... seems apt-get got wedged and was admant on "holding back" certain packages. with a little encouragement via adept it all went in, but this was harder than i've come to expect. which says a lot about how slick things are these days in general because a few years ago i would've been happy if that's how easy it was ;)
the sky2 driver is still broken, though. yay for that. and ipw2200 occasional loses its mind too. loverly. both known problems.
more on the amarok icons: they have a number of aesthetic and quality issues. not something i'd put as the default myself, and not just because breaking consistency like that is a fairly braindamaged thing to do. but i can see why they did it: many of the new icons are more consistent with each other, clearer in meaning (a great example is to compare "equalizer" from crystal and "equalizer" in amarok's set) and give the app much less of a "dancing rainbow of colours" look that one gets right now with crystal's myriad of colours. and it's not that the crystal icons are individually bad (generally they are higher quality than what i see in amarok's set, to be honest) but together as a collection of icons they really fall down. (btw, i really like the new "playing song" animation in the playlist; it's just that much nicer now)
all of these are things that oxygen is addressing: clearer meaning, non-busy (both colour and shape complexity) action icons, clarity between the icons as a set, etc.. so i hope that in kde4 amarok can drop their own icon set and use oxygen instead. i really think amarok can get what they need out of oxygen without resorting to the broken approach of having their own miniset. and if they really must, then fall back to a set of oxygen-y icons packaged with amarok if oxygen isn't the icon set used.
anyways ... the last few days have been a bit on the annoying side leading up to akademy. which sort of describes the last several weeks, too. things that have time pressure or community pressure keep coming up and landing at my feet and so i have to prioritize those items. what is nice is that there are more people stepping up again to do various things. i wonder if this is a summertime issue?
i also realized yesterday that i hadn't received my tickets for my flight to akademy. so i went to the website i bought them on (travelocity, if anyone cares) and the itinerary was there and said it was all good. so i phoned to find out what had happened, assuming the tickets simply had been misrouted. turns out my reservation had been cancelled. lovely. two hours later i have a new reservation that's nearly the same price i booked for originally on a better carrier. it wasn't easy to beat it into their callcenter staff that selling me a ticket they couldn't provide and then not updating the itinerary on the website was something they should take some responsibility for. *sigh*
and then i upgraded the stable kde on the laptop from the kubuntu repos and wow ... lots of little regressions. amarok crapped out a few times before finding its feet and boy are those new icons not so great. but more on that in a bit. i had problems with fonts, kontact and more after the upgrade ... seems apt-get got wedged and was admant on "holding back" certain packages. with a little encouragement via adept it all went in, but this was harder than i've come to expect. which says a lot about how slick things are these days in general because a few years ago i would've been happy if that's how easy it was ;)
the sky2 driver is still broken, though. yay for that. and ipw2200 occasional loses its mind too. loverly. both known problems.
more on the amarok icons: they have a number of aesthetic and quality issues. not something i'd put as the default myself, and not just because breaking consistency like that is a fairly braindamaged thing to do. but i can see why they did it: many of the new icons are more consistent with each other, clearer in meaning (a great example is to compare "equalizer" from crystal and "equalizer" in amarok's set) and give the app much less of a "dancing rainbow of colours" look that one gets right now with crystal's myriad of colours. and it's not that the crystal icons are individually bad (generally they are higher quality than what i see in amarok's set, to be honest) but together as a collection of icons they really fall down. (btw, i really like the new "playing song" animation in the playlist; it's just that much nicer now)
all of these are things that oxygen is addressing: clearer meaning, non-busy (both colour and shape complexity) action icons, clarity between the icons as a set, etc.. so i hope that in kde4 amarok can drop their own icon set and use oxygen instead. i really think amarok can get what they need out of oxygen without resorting to the broken approach of having their own miniset. and if they really must, then fall back to a set of oxygen-y icons packaged with amarok if oxygen isn't the icon set used.
anyways ... the last few days have been a bit on the annoying side leading up to akademy. which sort of describes the last several weeks, too. things that have time pressure or community pressure keep coming up and landing at my feet and so i have to prioritize those items. what is nice is that there are more people stepping up again to do various things. i wonder if this is a summertime issue?
Monday, September 11, 2006
qtimeline, timers and what to avoid in styles
played with qtimeline and animations over the weekend. qtimeline really takes out the annoying gruntwork of doing all the math for timers and what not out of the equation. one thing i did was make the new clear button in lineedits fade in/out when text was there or not, making it obvious that the button is associated with text being in there.
the animation bit was easy. the problems i ran into are that qlineedit::clear() isn't virtual anymore so an app can make the lineedit and the button fall out of sync by calling clear (though it'll fix itself once the user starts typing in it again) and there's probably not much i can do about it. cutting down on the virtuals was nice for startup time and what not, but it sure gets in the way when trying to make interesting things happen that weren't foreseen. which i seem to do fairly often. =(
the other issue i ran into was our icon theme not being svg. which means i only had crappy bitmaps to use for the animation if i wanted to use kiconloader to get the standard icons (which i do). the oxygen team is threatening to get oxygen into trunk over the akademy week at which point we need to add the ability for kiconloader to return svg data instead of a pixmap if desired. that way we can load the svg into an app and do things to it, like change the stylesheet or perform runtime compositions. the latter will be most useful for limiting the number of icons in a set while increasing the consistency; for instance, we could change the mimetype paper and all the mimetype icons would still match, even 3rd party ones.
luciano montanaro did some work on getting kicker to sleep more often and posted a patch for the clock to bugs.kde.org that makes it only trigger its timer once a minute if you aren't showing the seconds or blinking the dots. this being the default configuration, it will have a nice positive effect in that kicker will now go to sleep for a whole minute at a time instead of waking up twice a second. laptop batteries rejoice. i reworked the patch to sync to the minute more or less perfectly and check for drift so its still accurate to within a second or two and committed it for the next 3.5.x release.
and speaking of timers and kicker, i noticed while researching this a couple weeks ago that some styles do really nasty things that they just really shouldn't be doing. for instance, polyester starts qtimers when certain widgets appear (such as a combobox) and then never stops them. so as soon as a combobox shows, the app will wake up several times a second from that point forward. seeing as all apps use the same style and many/most use these common widgets, that means that most of the kde apps will end up waking up several times a second. this is really not nice for systems in swap or on battery power. ugh!
another fave of mine is qtcurve which changes the background mode on qtoolbuttons. this results in huge amounts of flicker in the taskbar, particularly after some of my recent painting optimizations where it just becomes more obvious. hint: don't change the properties of a widget in a style unless you can guarantee that it won't affect the painting of the widget. in this case, qtcurve should be checking that the widget hasn't set the background mode to nobackground. hopefully they'll fix this in the next release. but i won't get back the couple hours i spent trying to figure out what the hell was going on only to discover the offending code in qtcurve. =/
moral of the story is that when writing styles remember that its something that effects every widget and every app so one has to be very conservative and keep the potential side-effects "footprint" small.
i think the plastik and keramik developers, to name just two styles, have done well in this regard.
the animation bit was easy. the problems i ran into are that qlineedit::clear() isn't virtual anymore so an app can make the lineedit and the button fall out of sync by calling clear (though it'll fix itself once the user starts typing in it again) and there's probably not much i can do about it. cutting down on the virtuals was nice for startup time and what not, but it sure gets in the way when trying to make interesting things happen that weren't foreseen. which i seem to do fairly often. =(
the other issue i ran into was our icon theme not being svg. which means i only had crappy bitmaps to use for the animation if i wanted to use kiconloader to get the standard icons (which i do). the oxygen team is threatening to get oxygen into trunk over the akademy week at which point we need to add the ability for kiconloader to return svg data instead of a pixmap if desired. that way we can load the svg into an app and do things to it, like change the stylesheet or perform runtime compositions. the latter will be most useful for limiting the number of icons in a set while increasing the consistency; for instance, we could change the mimetype paper and all the mimetype icons would still match, even 3rd party ones.
luciano montanaro did some work on getting kicker to sleep more often and posted a patch for the clock to bugs.kde.org that makes it only trigger its timer once a minute if you aren't showing the seconds or blinking the dots. this being the default configuration, it will have a nice positive effect in that kicker will now go to sleep for a whole minute at a time instead of waking up twice a second. laptop batteries rejoice. i reworked the patch to sync to the minute more or less perfectly and check for drift so its still accurate to within a second or two and committed it for the next 3.5.x release.
and speaking of timers and kicker, i noticed while researching this a couple weeks ago that some styles do really nasty things that they just really shouldn't be doing. for instance, polyester starts qtimers when certain widgets appear (such as a combobox) and then never stops them. so as soon as a combobox shows, the app will wake up several times a second from that point forward. seeing as all apps use the same style and many/most use these common widgets, that means that most of the kde apps will end up waking up several times a second. this is really not nice for systems in swap or on battery power. ugh!
another fave of mine is qtcurve which changes the background mode on qtoolbuttons. this results in huge amounts of flicker in the taskbar, particularly after some of my recent painting optimizations where it just becomes more obvious. hint: don't change the properties of a widget in a style unless you can guarantee that it won't affect the painting of the widget. in this case, qtcurve should be checking that the widget hasn't set the background mode to nobackground. hopefully they'll fix this in the next release. but i won't get back the couple hours i spent trying to figure out what the hell was going on only to discover the offending code in qtcurve. =/
moral of the story is that when writing styles remember that its something that effects every widget and every app so one has to be very conservative and keep the potential side-effects "footprint" small.
i think the plastik and keramik developers, to name just two styles, have done well in this regard.
Tuesday, September 05, 2006
if you think user space sucks...
davej's talk on why user space sucks was really good. he'll be giving it again at akademy, so definitely check it out.
but to be fair to user space devs, it's not like the kernel devs are imbued with magic powers either. i've been having fun with linux on my laptop for two reasons beyond the usual "suspend working depends on the astrological chart of the day": the sky2 driver and partition checks. perhaps these are both problems with the ubuntu kernel only, but perhaps not ...
so, the sky2 driver was really broken when i first installed dapper on here. it often wouldn't even let me get an ip address over dhcp. it's slowly gotten better and now only dies every once in a while if there's sustained traffic in the 100-200k/s range. after one kernel update it would cause a kernel panic, but now it just stops working and i have to reload the driver. i understand the suse people have patches for this problem and they claim it works perfectly on opensuse. i wouldn't know because the machine i have opensuse on doesn't use the sky2 driver. but if this information is correct, why the hell isn't this patch shipping everywhere? thankfully wireless works like a champ so i'm only occasionally stuck with ethernet and the sky2 flakiness.
my favourite issue is with fsck on boot. currently an fsck is forced on every 30th power cycle. with a laptop, that happens fairly quickly. on a server that can and should be a long time between fsck's, and most desktops are probably similar.
but for someone travelling about turning the laptop off is often required, such as when getting on an airplane. given that suspend usually works, but not always, i also end up shutting down more than i'd like to. this means i get fsck's fairly regularly, which aren't exactly swift on a 100gb disk. i have two problems with this:
it's my computer. it's my file system. it's my data. software should never, ever, ever force something like an fsck on me unless the filesystem is having actual problems. preventative maintenance must be optional. why?
because maybe i need to boot up to find something in my email and i only have a couple of minutes to do so. the fsck comes up and i'm screwed.
or perhaps i need to look something up and i'm on battery and don't have the power left for a full fsck. i often drain my laptop down to <5% battery on the airplane, giving me enough juice for an emergency boot in the airport/cab/wherever in case i need to. an fsck would be very, very bad at this point. i was told by one kernel dev that the fsck only happens when there is power and i believed him. foolishness. =) for just the other day i powered up unplugged at a coffee shop and the fsck started. great way to waste my battery.
so if one can honestly say user space sucks (and it does ;) then at least we have good company when it comes to the linux kernel. =P
but to be fair to user space devs, it's not like the kernel devs are imbued with magic powers either. i've been having fun with linux on my laptop for two reasons beyond the usual "suspend working depends on the astrological chart of the day": the sky2 driver and partition checks. perhaps these are both problems with the ubuntu kernel only, but perhaps not ...
so, the sky2 driver was really broken when i first installed dapper on here. it often wouldn't even let me get an ip address over dhcp. it's slowly gotten better and now only dies every once in a while if there's sustained traffic in the 100-200k/s range. after one kernel update it would cause a kernel panic, but now it just stops working and i have to reload the driver. i understand the suse people have patches for this problem and they claim it works perfectly on opensuse. i wouldn't know because the machine i have opensuse on doesn't use the sky2 driver. but if this information is correct, why the hell isn't this patch shipping everywhere? thankfully wireless works like a champ so i'm only occasionally stuck with ethernet and the sky2 flakiness.
my favourite issue is with fsck on boot. currently an fsck is forced on every 30th power cycle. with a laptop, that happens fairly quickly. on a server that can and should be a long time between fsck's, and most desktops are probably similar.
but for someone travelling about turning the laptop off is often required, such as when getting on an airplane. given that suspend usually works, but not always, i also end up shutting down more than i'd like to. this means i get fsck's fairly regularly, which aren't exactly swift on a 100gb disk. i have two problems with this:
it's my computer. it's my file system. it's my data. software should never, ever, ever force something like an fsck on me unless the filesystem is having actual problems. preventative maintenance must be optional. why?
because maybe i need to boot up to find something in my email and i only have a couple of minutes to do so. the fsck comes up and i'm screwed.
or perhaps i need to look something up and i'm on battery and don't have the power left for a full fsck. i often drain my laptop down to <5% battery on the airplane, giving me enough juice for an emergency boot in the airport/cab/wherever in case i need to. an fsck would be very, very bad at this point. i was told by one kernel dev that the fsck only happens when there is power and i believed him. foolishness. =) for just the other day i powered up unplugged at a coffee shop and the fsck started. great way to waste my battery.
so if one can honestly say user space sucks (and it does ;) then at least we have good company when it comes to the linux kernel. =P
grade 1
today is the p-man's first day of grade one. it's a small school only 3 blocks from the house and so far i really like the teachers and what not.
they started by gathering everyone in the gym and students were called up one by one as they divided up the classes after a short introductory speech by the principle. they of course mispronounced the family name as "say-go" and so as p. walked towards his class group he paused when passing the principle and informed her in a polite but loud enough to be heard voice that it's "psy-go". he is so my son. =)
i'll post a pic or two later today (his mom has the camera atm).
they started by gathering everyone in the gym and students were called up one by one as they divided up the classes after a short introductory speech by the principle. they of course mispronounced the family name as "say-go" and so as p. walked towards his class group he paused when passing the principle and informed her in a polite but loud enough to be heard voice that it's "psy-go". he is so my son. =)
i'll post a pic or two later today (his mom has the camera atm).
Subscribe to:
Posts (Atom)
