No, you're confused. The point here is to have a hash proving the date you owned the original. So long as you don't actually lose all copies of the original file you can use other methods to compare the original and the claimed "copy" - it's the proof that you had the original file at a particular date that's important here...
Sorry - you had some way of changing how people are? Or were you wishful thinking that the universe worked differently than it actually does, because that tends not to work well in engineering, and funnily enough it doesn't work so well in marketing either...
Ads a few decades ago were somewhat more intelligent, they would actually talk about a product and list some specific reasons why it was better than another product.
Good for you 1%er that you prefer those ads. Now - I wonder why we don't see many of those any more... you'd almost think, I dunno, that they didn't work as effectively or something. Colour me shocked.
No, but then I don't tend to write applications that write files that I don't "own".
So - if I own the space where the file is going, then I figure that people who want a pointer to the most recent copy can use symlinks.
2) what?? No way. You resolve the symlink, and rename the target over itself. Assuming that you have permission to do that, otherwise you replace the symlink with the new file, and it's no longer a symlink, assuming your goal is to have the new file appear in the place.
Or - you fdatasync.
Or - you append the new data to the file rather than truncate, and finally update a record at the start of the file saying "valid content starts here and ends here" - eventually you have enough stale blocks at the start that you can overwrite them instead. Gosh, looks like we've invented a database format.
POSIX has no "atomic truncate and write new content to the same inode" operation. Denying reality and wishing it weren't so is a recipe for failure, the only difference is how often you see that failure.
Ah slashdot, where any ignoramus can have a strongly held opinion. Shit was written that took advantage of something that was specifically documented NOT TO BE THE RIGHT WAY OF DOING THINGS and just happened to be more safe due to conservative defaults in an earlier system. More modern filesystems came along tuned for more performance (guess what, you probably want that too) and suddenly the assumptions were less right.
Go mount with -o sync if you're so sure you know the purpose of a filesystem. Slow and safe boys, slow and safe. Enjoy your super reliable computing experience (especially KDE boot with lots of little file writes)
(as in: why the fuck are you rewriting a file with hardlinks in the manner? Use symlinks if you want "follow the changes" and hardlinks if you want "copy on write", easy)
He said he'd only give it to the Mayor, since nobody else was authorised to have that much power over the network. Sounds fair enough to me. And then he did hand the passwords over.
As plenty of other people have said, you really don't want to be handing out the passwords to a room full of people, especially if you don't trust those people.
Er, the world where they have full disk encryption enabled on their laptop, and they have a password to access it?
That's the sort of world where if you had encryption on the drive then a management interface that exposed a way to say "also encrypt all new keys so this public key has access".
I'm supposed to trust my crypto keys to a third party?
Man, it's all about you, isn't it.
This sort of shit is _really_ useful in a business, where the business security people have a master key that they can use to recover your data when you forget your password (yet again).
You might even have a friend or family member that you trust enough to keep a separate key that can read your data if you screw up your passwords or, y'know - DIE. So that your descendants don't lose everything you've ever done because it's locked behind personal encryption. Some legacy that brick is going to be.
Wasn't that their point? That it's a separate system, so that if it fails, you'll still at least have the command line?
Yeah, nice theory - except what actually happens is you have two different drivers both talking to the same hardware, and not talking to each other. The usual failure mode when it really does fail is to confuse things so much that the console driver doesn't work either.
Having X talk to the low level card management drive will mean that when X screws up, the console driver still knows what's going on and can keep driving the card.
When I installed Ubuntu for my parents I wound up writing a PPP connection manager that provides a controller applet in the task bar that can detect disconnected ISDN modem and other error conditions. It also has a web-based interface that allows them to connect other machines wirelessly or locally to the connection and just type in "internet" to a web browser to get the connection manager.
Honestly, with modern processors, SSL is barely a blip on the radar. We use postfix and nginx for our user facing connections, and I never see either set of processes using any significant amount of CPU.
You can upload >100Mb over FTP, but not over DAV - due to the way our proxy infrastructure works - there's a per-web-request size limit that we can't easily increase.
(Support via slashdot thread!)
Yeah - the DOS attack would be the major reason. Also, Cyrus doesn't support folder creation out of the box, but I suspect that would be a relatively easy thing to add. Still, DOS. We have an experimental "folder limit" patch that can be used to stop more than a certain number of folders being created for any one user... hmm. Sorry, still unlikely to be a huge priority for me:)
What I do is just check the "Lists" toplevel folder for folders that I should be creating, then search and move the initial set of messages. Given that I don't sign up to lists _that_ often, it's really quite a trivial amount of work.
Our last big outage was last time we had a 2TB single Cyrus instance. Do you know how long it takes to fsck 2TB on slow SATA RADI6? Neither did I until we did that. Over a week.
Basically, we lost 2 drives and had a third throwing errors within a few hours. It happens more often than you'd think. That was the last straw for the big partition idea.
We now have 300Gb data partitions, and Cyrus replication (check the mailing lists, I have written a ton of patches to Cyrus over the past couple of years - the only remaining issue with replication is it sometimes loses folder subscription information - but we run a weekly full replication check that notices that and fixes it)
It's not so much auto-focus as taking multiple pictures over time and eliminating the blur. Obviously slightly less useful for moving things than stationary things, unless you can define the movement accurately and input that into the algorithm (for example a car moving in a straight line at relatively constant speed)
I'm posting this not so much for you as for other people reading "automagic" and not understanding there's actually science for that magic:)
(what can a TLA do with an unlimited budget? I shudder to think. Probably waste (unlimited - delta) of it. What they do with the delta though, that's interesting.
Slashdot Mirror
No, you're confused. The point here is to have a hash proving the date you owned the original. So long as you don't actually lose all copies of the original file you can use other methods to compare the original and the claimed "copy" - it's the proof that you had the original file at a particular date that's important here...
Sorry - you had some way of changing how people are? Or were you wishful thinking that the universe worked differently than it actually does, because that tends not to work well in engineering, and funnily enough it doesn't work so well in marketing either...
Ads a few decades ago were somewhat more intelligent, they would actually talk about a product and list some specific reasons why it was better than another product.
Good for you 1%er that you prefer those ads. Now - I wonder why we don't see many of those any more... you'd almost think, I dunno, that they didn't work as effectively or something. Colour me shocked.
See my parent who was claiming that it's a filesystem limitation that you don't get sync behaviour.
Dur. That's slow, so you don't do that if you want speed - but don't come whining that you can't have your cake and eat it too.
No, but then I don't tend to write applications that write files that I don't "own".
So - if I own the space where the file is going, then I figure that people who want a pointer to the most recent copy can use symlinks.
2) what?? No way. You resolve the symlink, and rename the target over itself. Assuming that you have permission to do that, otherwise you replace the symlink with the new file, and it's no longer a symlink, assuming your goal is to have the new file appear in the place.
Or - you fdatasync.
Or - you append the new data to the file rather than truncate, and finally update a record at the start of the file saying "valid content starts here and ends here" - eventually you have enough stale blocks at the start that you can overwrite them instead. Gosh, looks like we've invented a database format.
POSIX has no "atomic truncate and write new content to the same inode" operation. Denying reality and wishing it weren't so is a recipe for failure, the only difference is how often you see that failure.
Let me know when you get that working in ext4 ;)
(seriously, I should have said "the POSIX filesystem API" to cover myself from pedants. Oh well)
You're welcome to write lots of little files. It will just be slow if you sync them all, or unsafe if you don't.
Same way a database will tell you to wrap lots of actions in a single transaction if you don't want the cost of a full commit after each action.
Except the filesystem API doesn't have any way to says "commit these 500 little files in a single transaction", unfortunately.
Annoyingly, it also doesn't have "unlink this directory and the files inside it in a single transaction", because unlink performance blows goats.
Ah slashdot, where any ignoramus can have a strongly held opinion. Shit was written that took advantage of something that was specifically documented NOT TO BE THE RIGHT WAY OF DOING THINGS and just happened to be more safe due to conservative defaults in an earlier system. More modern filesystems came along tuned for more performance (guess what, you probably want that too) and suddenly the assumptions were less right.
Go mount with -o sync if you're so sure you know the purpose of a filesystem. Slow and safe boys, slow and safe. Enjoy your super reliable computing experience (especially KDE boot with lots of little file writes)
[citation needed]
(as in: why the fuck are you rewriting a file with hardlinks in the manner? Use symlinks if you want "follow the changes" and hardlinks if you want "copy on write", easy)
mount -o sync. Enjoy your slow returns and strictly ordered writes.
He said he'd only give it to the Mayor, since nobody else was authorised to have that much power over the network. Sounds fair enough to me. And then he did hand the passwords over.
As plenty of other people have said, you really don't want to be handing out the passwords to a room full of people, especially if you don't trust those people.
Yeah, I filked that song with "pizzas" long time ago.
"Pizzas come in a van, they were put there by a man, at Mondo's Pizzaland, downtowwwwn"
Sang it at the start of the pizza sculling competition at college (woot, I was the returning champion, after all)
No, but you can have a comment if you like...
Er, the world where they have full disk encryption enabled on their laptop, and they have a password to access it?
That's the sort of world where if you had encryption on the drive then a management interface that exposed a way to say "also encrypt all new keys so this public key has access".
encryption key management applications...
I'm supposed to trust my crypto keys to a third party?
Man, it's all about you, isn't it.
This sort of shit is _really_ useful in a business, where the business security people have a master key that they can use to recover your data when you forget your password (yet again).
You might even have a friend or family member that you trust enough to keep a separate key that can read your data if you screw up your passwords or, y'know - DIE. So that your descendants don't lose everything you've ever done because it's locked behind personal encryption. Some legacy that brick is going to be.
Wasn't that their point? That it's a separate system, so that if it fails, you'll still at least have the command line?
Yeah, nice theory - except what actually happens is you have two different drivers both talking to the same hardware, and not talking to each other. The usual failure mode when it really does fail is to confuse things so much that the console driver doesn't work either.
Having X talk to the low level card management drive will mean that when X screws up, the console driver still knows what's going on and can keep driving the card.
Much better.
When I installed Ubuntu for my parents I wound up writing a PPP connection manager that provides a controller applet in the task bar that can detect disconnected ISDN modem and other error conditions. It also has a web-based interface that allows them to connect other machines wirelessly or locally to the connection and just type in "internet" to a web browser to get the connection manager.
All this means is, the human species will evolve in unison, rather than splintering off into different daughter species. How is this a bad thing?
http://en.wikipedia.org/wiki/Monoculture
http://www.answers.com/topic/heterosis-1
Honestly, with modern processors, SSL is barely a blip on the radar. We use postfix and nginx for our user facing connections, and I never see either set of processes using any significant amount of CPU.
You can upload >100Mb over FTP, but not over DAV - due to the way our proxy infrastructure works - there's a per-web-request size limit that we can't easily increase.
(Support via slashdot thread!)
Yeah - the DOS attack would be the major reason. Also, Cyrus doesn't support folder creation out of the box, but I suspect that would be a relatively easy thing to add. Still, DOS. We have an experimental "folder limit" patch that can be used to stop more than a certain number of folders being created for any one user... hmm. Sorry, still unlikely to be a huge priority for me :)
What I do is just check the "Lists" toplevel folder for folders that I should be creating, then search and move the initial set of messages. Given that I don't sign up to lists _that_ often, it's really quite a trivial amount of work.
http://www.fastmail.fm/pages/fastmail/docs/reliability.html
Our last big outage was last time we had a 2TB single Cyrus instance. Do you know how long it takes to fsck 2TB on slow SATA RADI6? Neither did I until we did that. Over a week.
Basically, we lost 2 drives and had a third throwing errors within a few hours. It happens more often than you'd think. That was the last straw for the big partition idea.
We now have 300Gb data partitions, and Cyrus replication (check the mailing lists, I have written a ton of patches to Cyrus over the past couple of years - the only remaining issue with replication is it sometimes loses folder subscription information - but we run a weekly full replication check that notices that and fixes it)
And yeah - our staff read slashdot!
I may be, I may be, but your grandmother smells of elderberries, so there.
You count funny
The Cold War is coming back at a very bad time for the US.
Well, duh. Since when did anyone bring war to someone for whom it's a _good_ time?
It's not so much auto-focus as taking multiple pictures over time and eliminating the blur. Obviously slightly less useful for moving things than stationary things, unless you can define the movement accurately and input that into the algorithm (for example a car moving in a straight line at relatively constant speed)
I'm posting this not so much for you as for other people reading "automagic" and not understanding there's actually science for that magic :)
(what can a TLA do with an unlimited budget? I shudder to think. Probably waste (unlimited - delta) of it. What they do with the delta though, that's interesting.