Understanding is "horizonal" - there's always, always further, beyond wherever you reach. Thus, there's almost never the luxury of understanding everything that's relevant. Instead, you can only identify what may be sufficient for your purposes, and aim to understand enough for adequate confidence about the risks.

that may have been intended as stinging sarcasm, but the problem is with a component of HTC's proprietary Sense overlay. that sorta takes any point out of your mockery.

well said.

lisp was my favorite language until i found python. i still like lisp a lot - i develop some emacs extensions, like allout, and enjoy delving into the code. i'm also a big shell scripting fan, and i even love scheme's compactness. however, whenever i return to python from any other computer language, i am struck by its comprehensibility.

it's unfortunate, but "comprehensibility" is hard to pin down, and critically important. particularly in dynamic situations - which is when scripting is most needed. the more that a language has syntactic clutter, and/or is excessively naive (so a lot of code needs to be written and rewritten for simple things) or sophisticated (requiring the programmer to be so clever that they and others inevitably have a hard time understanding what they've written), the harder it is to evolve a system. of any language i've seen, python strikes the best balance, by far, in all of those realms.

it's kinda whacky that the GNU folks have stuck with the choice of guile after all this time, and in the face of others leaving that quest. witness MIT migrating from scheme to python for their introductory programming courses...

• Debugging is twice as hard as writing the code in the first place. Therefore, if you write the code as cleverly as possible, you are, by definition, not smart enough to debug it. -- Brian Kernighan

unlinking, and relinking when necessary, is a good suggestion. it's nice to have a useful workaround out of a slashdot discussion!

(no, i'm not being snarky - i hadn't even looked at the unlink option since i first started using dropbox, since i never use it on any but my own computers.)

thanks!

(it's also amazing to me that they don't use SSH / SSL.)

(i see you've fallen for the whole "tinfoil hat protects your brain pan" scheme. any pastafarian knows that only lightly cooked spaghetti - al dente - can protect your noodle without disastrous side effects.)

your indignation suggests unwarranted faith in techniques that the article's authors acknowledge could be easily fooled (they say, "gamed"), since they trace only data flow, not control flow - and that is all they can reasonably do. so all that android-instituted behavioral tests would achieve is another level of misplaced trust. great goal.

really, android's existing sandboxing scheme is much more worthy of ongoing trust than the described profiling scheme. the sandboxing scheme cannot claim the granularity of the profiling scheme, but it realiably covers what it claims to cover. the profiling scheme cannot.

well said - but for me, at least, the "icing" part underestimates the value of openness.

i'm near completing my second year of owning an android G1, and the thing has been spectacularly useful things i've owned. somewhere near the beginning of my second year i was increasingly frustrated with the limited apps storage space, though, and general thrashing of the android 1.6 install (perhaps due to my crowding it with apps, but i was trying to cut out unnecessary stuff, honest). rooting with cyanogenmod became pretty easy, and enabled me to use part of my 8 GB card (now 16GB:) for app storage. i'm now running android 2.1, thanks to cyanogen, and the phone is working better than it ever did at stock 1.5 or 1.6. it continues to be spectacularly useful 2 years out, where it was running out of steam at the end of one year while sticking with the stock system.

mind you, android 2 is not likely to ever be released for my phone by my vendor, and it would have been increasingly untenable for me to stick with this phone anywhere near as long as i have - despite loving the format (really decent physical keyboard, generally decent other stuff), and not seeing satisfying alternatives. for those reasons and others i could see staying with this thing for a while more - and if htc had locked out alternative os loads, i would not have had the choice.

consequently, as far as i can tell motorola phones, with their signed-boot restrictions, offer no lasting value to me, whatsoever, and are not in the running. i'll be eventually be looking for another android-based phone with a hardware keyboard - but it doesn't have to be soon, and it definitely won't be one that limits the long-term utility of the thing so drastically.

exactly what is useful can sometimes be difficult to gauge in advance, and that's where openness - allowing more options over the life of even an appliance - can be worth a lot.

beautiful story - thanks!

oy vey.

others have responded better than i can, except i see one aspect missing from the other comments. google's action signals a significant shift in the way it relates to what amounts to an intransigent bully in charge of the biggest ball game (economy) in town.

prior behavior (by google and every other big player) has been mostly accommodation - walking gently around the bully in the hopes of staying in good graces, to not get shut-out from the game. google's recent actions - effectively accusing the govt. of the corporate spying, dropping the google.cn censorship, etc. - amounts to shrugging its shoulders and saying, "playing by your rules is not worth the grief". the way i see it, clinton's response is a signal that google's is not just a unilateral, empty gesture - this is policy shift is endorsed by both.

the departure is not from engagement, but rather from continuing to unequivocally accept china's rules of the game.

in order to have any real significance, google actually has to be willing to turn down some huge opportunities. they've demonstrated willingness to commit to big stakes in, eg, the way they influenced the terms of the spectrum auction, so i see no reason to view any of this as feeble gestures, but rather a genuine shift. interesting.

anyone interested in this distinction might appreciate the model described in finite and infinite games by james p. carse. it's a kind of convolution of the tao te ching, distilled down to:

A finite game is played for the purpose of winning. An infinite game is played for the purpose of continuing the play.

carse might say that performance-orientated people (paul) are occupied with the resulting claim - title, status, accomplishment, authority, etc - that they can make looking back on the win. those that are mastery-oriented (matt) are more concerned with developing ability to continue the play into the ("horizonal" - always in the advancing distance) future.

Surprise causes finite play to end; it is the reason for infinite play to continue.

and

To be serious is to press for a specified conclusion. To be playful is to allow for possibility, whatever the cost to oneself.

and

Because infinite players prepare themselves to be surprised by the future, they play in complete openness. It is not an openness as in candor, but an openness as in vulnerability. It is not a matter of exposing one's unchanging identity, the true self that has always been, but a way of exposing one's ceaseless growth, the dynamic self that is yet to be. The infinite player does not expect to only be amused by surprise, but to be transformed by it, for surprise does not alter some abstract past, but [by discovery of what actually happened,] one's own personal past.

(i wonder whether those interested in this kind of topic would more tend towards the mastery/infinite-play perspective?)

anyway, one of the most illuminating books i have read, along with the tao te ching (and, the one other on my paltry list, the politics of experience by r. d. laing).

ken

hmm. i'm running more in the background now than i was before, because there are some apps which have background jobs (eg, google voice, selfhelp) which i couldn't afford to have on the phone at all, space wise. no problems now, except for some problematic apps (eg, where).

i would often try killing background tasks (using advanced task manager, very nice), but never got signficant improvement - to the point of killing all.

(on second thought, i suppose the eeprom must be used, to preserve dynamic parts of the filesystem across reboots. in any case, the phone performance gets awful as apps increase.)

i don't think so, but i could be mistaken.

the specs are 192 MB RAM and 256 MB EEPROM. i believe that the eeprom holds the firmware loader and static OS elements, while the RAM is used for active operating memory and varying elements of the internal-phone filesystem, including apk storage, data caches for things like gmail and the browser, and so on.

i can tell you that the space available for the .apks starts at something like 75 MB total. that's quickly consumed (with some substantial apps taking several MB apiece), and i experienced awful performance problems even though i left 12 to 17 MB (depending on how recently i cleared the caches).

i understand the bind that google/android is up against, and think it is terrible both in principle and in personal impact.

in order to limit copy-access to android app executables, android depends on sequestering apps in phone storage. while most app producers don't care about limiting access to their executables (apk's), some commercial vendors do. (some common evidence of this is the way that most apps are available for copying by android backup programs like MyBackup Pro, but some aren't.) of course, root access defeats this sequestering - and, in fact, the biggest performance advantage on machines like my G1 is due to jiggering things, with symlinks, etc, so that app storage (as well as some resource cache storage) is physically on the SD card.

the terrible bind is that, on phones like the G1, the phone-storage RAM (192 MB) is a critical resource shared across operational and storage functions, so that the phone works terribly if you have too many apps. and "too many", for a phone that's supposed to be very multi-purpose and extensible, is disturbingly few. it really is a fatal flaw - until i upgraded to cyanogen's mod, things like scrolling would fail to respond most of the time, returning to the home screen or starting an app could take on the order of minutes, etc. and this after i removed a lot of apps, including ones that were occasionally crucial. after upgrading to cyanogen the device works like an, um, dream. i can run everything i need, and more, and the phone is sliced-bread-caliber useful with quick, smooth responsiveness. happy dance!-) now they're bringing down the boot on my savior. darn.

it seems obvious to me that google can't afford to allow undermining of their key provision for proprietary vendors who don't want their .apk's loose in the wild. it's a platform-policy agreement they made. it seems equally obvious to me that this is a damn shame - a profound architectural restriction solely for the purpose of a few overly restrictive vendors, who also happen to be some of the big vendors.

(from many comments, elsewhere, about similar relief from upgrading to cyanogen, i see that my experience with the G1 is not unusual. going back to the standard android release is not an option, so figure i'll stick with my cyanogen install until my contract is up, sometime early next year, and by then there should be other android devices with a physical keyboard and without the cripplingly insufficient amount of RAM. i truly am sad that google is in this bind, and feel that the current arrangement for securing apk's is profoundly flawed, and finding a different approach deserves substantial effort.)

ken