« first day (1139 days earlier)      last day (4174 days later) » 

Bob
Bob
16:00
@Hennes normallt safer to look at value/worst
Can you try to read from that disk and see if that number increases?
Bob
Bob
but they don't look that bad
@Hennes Copy other files? Or just open them?
uhm, I would just boot BSD or linux and dd to dev null
Or use windows checkdisk on the whole disk (surface scan)
I'm just from linux on the pc with that hdd..
16:02
MY windows knowledge is limited here. I tend to remove problem disks from laptops and plug them into a linux or BSD desktop to diagnose
Bob
Bob
@Hennes before doing either of those, be prepared to lose all data
in other words, make sure everything you want has already been recovered
with failing drives, ddrescue is great
in that case either dd if=/dev/sda of=/dev/zero bs=64k or ddrescue
Bob
Bob
@Hennes chkdsk can make volumes unreadable sometimes :\
Again, before trying anything like that, make surey ou have a copy of everything you need to keep
the first command reads the entore disk (make sure you specify the right value, sda for the first disk, sdb for the second) and ignores the results. The second command is usueful for recovering data from a failing drive and will not abort when it runs into an error
Aye
Bob
Bob
a surface scan is great for triggering/detecting errors
16:03
Yes.. I'm almost finished the backup..
Bob
Bob
but that also means it's great at permanently losing data when the errors trigger
Step 1) Check your backups
Step 2) Make a new backup if needed
Step 3) Try to fix
Bob
Bob
basically, you have a bunch of 'maybe-error' sectors, that might be revealed to be fine or might be triggered into full 'definitely-error's (which could potentially lose data - often, a failing drive can only be read once if at all and it's stupid to waste that single read on error detection when you could have recovered data with it)
this will happen on the next read (hence surface scan)
Step 4) Create automated incremental backups so you never have to worry about Steps 1 and 2
That would be ideal. Esp. if made to an off-site location and even better when taken off-line
But lets not get ahead of ourselves
16:06
@Hennes This is why crashplan is freaking amazing :)
Bob
Bob
@DarthAndroid if you can use it -_-
@Bob Country restrictions?
Bob
Bob
I really need to figure out some form of offsite backup
@DarthAndroid upstream is too shit
80kB/s max
and too much data
They do heavy compression and deduplication before sending it
My parents aren't on a good connection either
Bob
Bob
I calculated at some point that it would take me a full decade at full speed
16:07
Crashplan's backups are encrypted before they're sent out. Ah, that is what I required.
@DarthAndroid Some advices? I just used free file sync to backup my data.. but never understood properly how to obtain an incremental one..
Bob
Bob
@DarthAndroid I have over 4 TB of data now, including quite a fair bit of incompressible data.
Not the best situation :P
rsync
rsync is great for incremental backups
@Overflowh I generally recommend crashplan these days. You can use their software for free if you have your own place to store the data, or they provide a very reasonable online server to store stuff offsite
I just does not like windows permissions
16:09
I think I'm paying $4/month for 15-minute incremental backups of my linux media server, windows desktop, mac laptop, my mother's desktop and laptop, sister's laptop, dad's laptop, and other sister's laptop. About 10TB worth of data currently.
As the tech guy that gets the call when stuff doesn't work in my family, it's so very satisfying to know that even my sisters can't screw something up where they lose data.
I need to look more careful at that crashplan thing.
Bob
Bob
@Hennes At least it's a default ACL implementation.
ACLs on Linux can be painful to set up.
Though I will nt use non-opensource stuff to archive my password files
@DarthAndroid Oh, I read about crashplan. It would be a good thing if my upload was decent..
Or my dairy
16:11
@Overflowh Yeah, the upload can bite you, but with my parents I just capped it at a % of their total upload and then just let it run for... 4 months is it now?
took forever, but it eventually got all their photos and movies backed up
Bob
Bob
@DarthAndroid and what is this speed? :P
@Hennes A really intelligent move. But it would be better to remember the passwords, not to store them..
If I only could
Some of them are 32 char of random data
I think I had it set at 128KB/s or so.
With a single unique password per account
Create password: step 1: mash the keyboard
Bob
Bob
16:13
@DarthAndroid And that's a portion of their total -_-
[ev'5jh;3 ljh;dtk; nk
Step 2: Add some weirdness
E.g. change it to [eGv'5j[(JJ.`#;3 $ljh;dtk
Bob
Bob
That's closer to what I'd get at a speed that allows me to use my connection more or less normally for other purposes.
@DarthAndroid Well, I think it's impossible. My upload is, officially, 386kbps, but really it rarely pass the 35kbps.. At that speed, it makes my connection useless.. and taking just a % of it, it would require years to upload everything..
step 3: curse at the website stating that the password is not complex enough
Bob
Bob
A little over 3 years.
Better than the last time I calculated, actually. Must've forgotten something.
16:15
E.g. dells website told me it needed to contain a number. They mean "We only check the first 8 chars of your password for numbers. Yours are past char 9"
Bob
Bob
kbits or kBytes?
@Hennes Oh.. well, so keep using opensource stuff for passwords :)
@Bob @Overflowh Well, Crash plan's software can still be of use. It also supports backing up to local LAN boxes, and removable harddrives. My desktop actually backs up to both CrashPlan's off-site servers and my local media server
BTW, what is the g for in gb. Girly bits?
Bob
Bob
@Hennes that or you're using filtered characters (SQL injection worries? :P)
16:16
SQl injection is so 1990.
if you're not using their online storage service, the software is free
Bob
Bob
@DarthAndroid I'm already using Acronis True Image - just trying to figure out a viable offsite method :P
The last time I needed that was almost a year ago
@Bob Sneakernet :)
Hand a few harddrives to your neighbor down the street, offer to hold theirs for them.
Bob
Bob
Eh.
Then it would have to be encrypted, and would take a dreadfully long time to update.
16:18
@Hennes It would be a bad idea to create a small tool that generates and stores passwords for you?
Use crashplan to fill the external drives.
It's encrypted and incremental.
anyone else running into an infinite redirect loop at msi.com ?
@Bob probably comes down to it being an unlockable device and Verizon wants everything on their network locked down
@Sathya Nope.
16:19
what about in.msi.com ?
Bob
Bob
@allquixotic Which is illegal, no?
@Sathya would SU want this question?
@Sathya Immediately redirects to us.msi.com for me
@DarthAndroid This is the think that captured my attention earlier.. I will surely give a try to this crashplan software.. Actually, I back up things on my external hard drive, but I'm quite messy
Bob
Bob
@Overflowh KeePass already exists :P
@DarthAndroid au.msi.com here
16:20
(oh hey @pops, congrats on the new job)
@DarthAndroid so some screw up at the Indian site.
@Overflowh An eWallet? No, that would be a lot better than an encrypted file.
@Sathya that would be my guess
for me msi.com -> redirect to in.msi.com -> redirect -> msi.com times infinity
And it already exists. Just need to start using it
@Sathya That one doesn't work for me.
Bob
Bob
16:21
sigh retarded chat flags again
Is "45 minutes until high as a kite" a flaggable thing?
Bob
Bob
@Hennes I don't see why it would be :S
@Pops hmm, send it over.
Hmm
Bob
Bob
and it actually got removed.
wow.
16:21
@Bob I know, also lastpass and many others.. But I have a problem trusting those tools.. I still remember all my important passwords in.. my brain..
@Hennes eWallet? Never heard before
It is the name used for a program which safely stores your passwords
Bob
Bob
@Overflowh KeePass is fully local and open source (and very commonly used).
Woa. 5 flag warning in 1 minutes
Bob
Bob
16:23
Not sure if anyone has actually audited it, though.
What kind of room is 'the bridge' thouigh
@Hennes And is it opensource? Are you sure that the code and the storing method cannot be bruteforced?
@Sathya Wow, you even saved me the work of editing. Alrighty!
@Sathya And thanks!
@Overflowh It's open source, and I haven't personally checked, but it's popular enough that I feel comfortable that enough people that have the free time have looked over it.
I still just stick with my PwdHash stuff though :)
Bob
Bob
@Hennes A weird and mysterious place.
16:24
@Pops cheers!
@DarthAndroid This is a good reason :P
Seems to be a gaming room
Anyway, I think I will use my brain until it will collaborate :P when it will start giving me problems like the hard disk of my laptop, I will use keepass :P
Bob
Bob
@Hennes Isn't that just '5 people agreed'?
which is pretty ridiculous for that flag, too
Meh, I sort of agreed. It is spam, but the other kind of spam.
Just ask the person to cool down and stop
16:27
HAMMERTIME!
@Bob However.. about the wolframalpha link you posted before.. I'm quite sure that they aren't KB/s, but Kb/s.. KiloBits per seconds..
/me hammers John Anvil
Lol
A rare example from the Belgian surrealist school of performance artist estate agents, who work wearing only their underpants.
Bob
Bob
@Hennes not quite 'ban-from-chat-worthy', unless they are being unreasonable
Aye
Bob
Bob
My connection is nominally 1Mbps up, practically 80kB/s up and 40kB/s up is the safe upstream without seriously affecting anything else.
16:29
Enough to drop in and look what is going on. Maybe ask them to spam a tad less
@Bob Wow.. mine is in Kb :/
Bob
Bob
@Overflowh :(
@Overflowh Is that the best speed from your provider at your location? Where are you located?
16:36
@DarthAndroid No well, not the best speed. If I lived at 34 kilometers from here, I could access the optical fiber, and they say that can be 100 MB in download and 10 in upload.
@DarthAndroid I'm from Italy. South of Italy.
Ah. May I ask roughly how much you pay per month for that connection?
@Bob @JourneymanGeek @DarthAndroid this looks cool osv.io
> OSv runs off-the-shelve JVM today. We plan to change the JVM by exposing OS internals and direct access for page tables, scheduling decisions and the raw IO layer
@DarthAndroid Mmmh.. about 40€ each two months.. that are about 53 USD
sounds like it's 100% secure... I mean, JVMs never have vulnerabilities that allow users to inject code, right? imagine the havoc they could wreak with kernel access directly :D
@allquixotic What are you talking about? The JVM does nothing but run user code. I tell it to run code, and it runs it. That's what the JVM does.
Bob
Bob
16:40
@allquixotic hm. you'll have to remind me to try that when I get my laptop back
Or are you talking about the abomination of a plugin that tries to support java in the web browser?
Bob
Bob
@allquixotic Wait, that much focus on Java?
No thanks...
@DarthAndroid No, I'm talking about osv.io
read the actual page
Bob
Bob
@allquixotic I love how it's using the default Joomla favicon
I'm talking about the potential exploitability of a remote code vulnerability in a JVM on OSv that happens to run in-process with the kernel
or actually as a kernel "task" (but with no context switch)
Bob
Bob
16:42
@allquixotic I thought OSv was supposed to be an 'OS' on top of some other virtualisation stack?
Where the virt stack will take care of isolation?
> JavaScript, Scala, Clojure and more run on top the JVM.
@Bob it is -- you make it a KVM guest (or whatever full hypervisor)
Bob
Bob
wait what... JS... on the JVM?!
@Bob oh yes... and it should be as fast or faster than V8 when it's done in JDK8
@Bob JS has been on the JVM for years, via Rhino
DarthAndroid was actually telling me about that earlier
16:43
The new Nashorn engine in JDK8 is supposed to be fast.
Bob
Bob
@allquixotic not sure if being sarcastic
@Bob not being sarcastic!
@Bob Not sarcastic.
V8's been doing JIT for a few years. JVM has been doing JIT for what, a decade? decade and a half?
fact: Oracle's Java HotSpot JVM has an extremely performant, stable, scalable multithreaded garbage collector and an extremely optimized JIT for bytecode, both of which are years more mature than the JIT in web browsers' JavaScript engines
now... take Java Script and run it through that JIT and clean up behind it using Java's GC
There aren't solid benchmarks yet, but the few preliminary reports are very, very exciting in terms of what the JVM is capable of with compiled javascript.
There's a reason that JPython was several orders of magnitude faster than the C implementation of python for a while.
16:47
JRuby ain't bad either
(lately though, CPython has closed the gap)
(And neither can touch Pypy.)
in about 2010 one of my professors told me we have 30 years of theoretical computer science research left to do before we've discovered most of the significant, relevant and useful optimizations that can be done in an efficient manner for JIT runtimes of this sort
there's a lot of upside left
and that's taking into account what Java, LLVM, etc. can do today
Also, all the new Lambda stuff in JDK8 is refreshingly nice. Very late, but nice.
there are also some theoretically known, discovered techniques that can be done in JIT optimization passes, but which are not normally done today because, on modern CPUs with typical code sizes, it would delay program startup by too long, so we either have to wait until CPUs get faster, or figure out how to work in those optimizations while the app is running without disturbing it
or someone needs to come out with a sort of "compile-time JIT" that will run the program with typical workloads, profile it, determine all possible optimizations (even the very very slow to produce / expensive ones), optimize it all to hell, then save the machine code binaries out to disk
static compilers don't actually run your code so they're very limited in what they can do
@allquixotic That's part of the goal of the -server flag. Doesn't save anything for later, but "I want you to not worry about startup time and do in-depth analysis of this code before running it."
Bob
Bob
16:56
@allquixotic That's what happens with native image generation with .NET, or AOT compilation with mono
though I understand that it has some downsides compared to JIT (can't optimise for different/changing workloads)
I think that's part of the idea though. Static / AOT compilers don' t know how the program is being used, and can't optimize fully.
Profile-guided optimization (PGO, sometimes pronounced as “pogo”) is a compiler optimization technique in computer programming to improve program runtime performance. In contrast to traditional optimization techniques that solely use the source code, PGO uses the results of test runs of the instrumented program to optimize the final generated code. The compiler is used to access data from a sample run of the program across a representative input set. The data indicates which areas of the program are executed more frequently, and which areas are executed less frequently. All optimizations be...
Intel's freeware, proprietary x86/x86_64 compiler does PGO for C/C++
that's the closest that non-VM languages come to the types of optimizations that JIT does
Bob
Bob
@DarthAndroid Puts far too much trust in the user. "Hey, the first time you run this, pretend you're running everything in the average case so we can optimise it."

« first day (1139 days earlier)      last day (4174 days later) »