/dev/chat

Gilles

00:20

@terdon Why did you convert unix.stackexchange.com/a/198998 to a comment? It's an answer. A wrong answer, but it's an answer, not a comment. That makes the comments under the question confusing now.

terdon

@Gilles 'Twas flagged as "not an answer" and it read like a comment.

I take it your comments on the answer were before it was edited?

Gilles

The first one was before the edit, the second after

terdon

OK, I didn't look closely enough and read them to mean that the answer was not useful.

I undeleted it.

Gilles

It's not useful because it's wrong (whatever x-gvfs-hide does, it's not that). But it's an answer. It isn't a request for clarification, or a hint, or a side remark.

Huh, did somebody upvote?

terdon

@Gilles I often convert too-short answers to comments. Probably a bad idea in that case.

Braiam

00:24

@Gilles you want only the wake up to stop or showing it altogether

Gilles

@terdon You shouldn't. You should convert answers to a comment only if they are not answers but are worth having for some other reasons. Not just because they're short.

@Braiam I want the wake up to stop. I don't mind the showing, but I figure if it's shown, then the application will try to read something.

terdon

@Gilles I don't convert just because they're short! I will convert something that might be useful but is not really giving an answer.

Gilles

And I hope it goes without saying (famous last words on the Internet), but unplugging the disk and the like are not acceptable solutions

terdon

Man, you're so picky!

Braiam

damn, I was about to suggest that!

terdon

00:27

::P

Braiam

01:01

@Gilles if you find a way to modify the udev database and set UDISKS_PRESENTATION_HIDE to 1, I think you could. Forget about gvfs and udisk

derobert

15:22

0

A: Prevent the Gtk file dialog from listing mount points

The GVFS documentation has a file about Controlling What is Shown in the User Interface. In short, you have two ways to do this: If it's in /etc/fstab, add x-gvfs-hide as one of the options (or, for older versions of udisks2, comment=gvfs-hide). Configure udev to set the $ENV{UDISKS_IGNORE}="1"...

... not every day you get to answer a question from @Gilles.

@Braiam that's udisks1. udisks2 uses a different variable

Braiam

@derobert dammed udisk manual

casey

15:41

time to file a bunch of bug reports! (not looking forward to that, but for the greater good I suppose)

derobert

@casey did you just upgrade to jessie?

casey

@derobert no, I went crazy with gcc 5.1 and now have a (small) list of packages that no longer compile with link-time-optimization (regression from 4.9) and another small list of packages that are broken from changes to the C pre-processor in 5.x.

the no-longer-compile-with-LTO list are all ICE cases, so those go to gcc, the rest go to their respective devs

the rest of my problems are all C++ ABI related, but those are nobodys fault but my own

but unrelated, yes I did upgrade my digital ocean VPS to jessie with no problems of note

derobert

@casey Well, you or the C++ committee :-/

casey

15:56

@derobert perhaps the gcc folks but I can't blame them. A major version bump is a good as time as any to break things. That said, gcc 5.x can emit/link objects against the last ABI but since I have a source distro on this box it doesn't make sense (to me) to mix and match and instead bring all the C++ stuff up to the new ABI

my complaint there is against gentoo actually, as the easiest way to rebuild the system doesn't take dependency order into account when ordering the package builds, so it doesn't handle ABI breakage well

derobert

I believe the ABI change was forced by C++ standard changes

@casey fun :-(

Darth_Vader

morning.. anyone know a good guide to learn what to be to pass a bash conditional and what [ -eq 1 ] means?

casey

@derobert probably, as the new ABI is called the C++11 ABI by gcc

@Darth_Vader [ is test, see man test

derobert

@Darth_Vader [ -eq 1 ] is an error... -eq is a binary operator, it needs a number on both sides

casey

or see the bash builtin for test

derobert

15:59

help test

as for a guide, you just want one on shell scripting. Any shell scripting guide is going to cover test/[.

Darth_Vader

ah was trying to figure out what [ $? -eq 1 ] meant

casey

$? is the return value of the last command run

so that is testing that the last command returned 1

Faheem Mitha

18:10

I'm playing with my new landline phone. The 3G network seems very slow. Is that normal? Of course, I can't tell if it is actually using a 3G network.

Hmm, calibre is trying to talk to my phone (currently plugged in and apparently recharging on USB).

Failed to get storage info for device.

Traceback (most recent call last):
File "/usr/lib/calibre/calibre/gui2/device.py", line 86, in run
self.result = self.func(*self.args, **self.kwargs)
File "/usr/lib/calibre/calibre/gui2/device.py", line 467, in _get_device_information
fs = self.device.free_space()
File "/usr/lib/calibre/calibre/devices/mtp/base.py", line 24, in synchronizer
return func(self, *args, **kwargs)
File "/usr/lib/calibre/calibre/devices/mtp/unix/driver.py", line 308, in free_space

No idea what this means, and not even sure why calibre is paying any attention to a phone.

casey

19:04

landline with 3G?

@FaheemMitha mtp is a protocol to sync multimedia files with phones (music, ebooks, videos, etc)

Faheem Mitha

@casey Oh. Not sure what to do with that, if anything.

Another question: what are best practices for recharging Android phones, if anything? Do I let it completely run down before recharging, or does it not matter?

casey

@FaheemMitha your log suggests calibre was trying to synchronize content to your phone via mtp.

Faheem Mitha

@casey And presumably mtp was not enabled on the phone side?

casey

android has used mtp rather than usb mass storage since 4.0

Faheem Mitha

@casey Oh. Does that mean it is enabled?

derobert

19:11

@FaheemMitha it's a setting, but its probably enabled by default

Faheem Mitha

@derobert ok.

derobert

should give you a notification when you plug it in telling you if its in MTP mode

casey

@FaheemMitha yes, it is likely enabled by default and that is probably why calibre tried to talk to it

Faheem Mitha

@casey but failed, apparently.

I just plugged it in. Haven't tried doing much. The 3G network doesn't seem to work inside my room.

derobert

I think the recommendation for lithium batteries is that it's better not to do full discharges, except possibly once a month or two to recalibrate

Faheem Mitha

19:12

I guess it is too closed.

@derobert Ok. How low should I let it get? And are less frequent charges better than more frequent?

derobert

should be an icon up top telling you what cell network its on. Or in settings/about

casey

Personally I try to not let my LiON batteries go below 12% and then I charge to full (I put on the charger at night regardless of battery level)

whether that is best practice or not, can't tell you

Faheem Mitha

@casey ok

derobert

@FaheemMitha shallow cycles are better, within reason. Also don't have it live on the charger.

Faheem Mitha

@derobert yes there was some 3G icon. But when I go into my room it goes dark (I think). I'm not really familar with how this phone works. BTW, what do I need to do to get a shell?

casey

19:14

yes, leaving on the charger needlessly is a good way to kill the battery much faster than normal

derobert

install a terminal app, or install adb and do adb shell over usb

Faheem Mitha

@derobert shalllow cycles means more frequent recharges? It was at 40% just now. It's taking its time recharging.

@derobert Android terminal app? I guess I should try to enable wifi off my router then.

derobert

@FaheemMitha it'll charge slowly from a computer, wall charger should be much faster. Yeah, shallow means more frequent charges

Faheem Mitha

Any recommendations for an Android terminal app? And does the underlying structure resemble unix?

casey

you should do that anyway, unless you have an unlimited data plan there is no good reason to use the cell data network at home

derobert

19:16

@casey I've been trying to use all 7G of data t-mobile gives me now. Haven't managed yet...

Faheem Mitha

@derobert Ok. There is what seems to be a wall charger in there, but I seem to be missing a lead. Is that normal?

@casey sure. though this router is kinda crap. I'll try.

derobert

probably takes a standard micro-usb cable, like the one you're using to connect it to the computer

casey

@derobert yea, if you have a good plan it doesn't matter, but that isn't universal

Faheem Mitha

@derobert Ok, maybe it plugs into the end of that cable. I'll take a look. Thanks.

@derobert Do you get it bundled?

casey

@FaheemMitha its a linux kernel with a non-standard filesystem layout, app isolation via unique users per app

if you want familiar tools, install busybox or a gnu userland

derobert

19:17

@FaheemMitha get what bundled? the data is part of my monthly plan with them

Faheem Mitha

@casey yuck.

casey

and if you need root you'll have to jump through some hoops

Faheem Mitha

@derobert the data.

derobert

I use Better Terminal Emulator Pro, but that's not free. There is something like Android Terminal which is also pretty good.

Faheem Mitha

@casey hmm. I'll let it be, probably. Don't really see much point.

casey

19:18

and if you just want to ssh out of your phone, get juicessh or similar

Faheem Mitha

It's just a phone. I might also use it as a camera.

I had a nice digital camera. It died shortly after arriving in India. Really very nice digital camera.

Even the comparatively low-end Android phone I have seems to take quite nice photos.

derobert

play.google.com/store/apps/details?id=jackpal.androidterm ... I think it's that one

@casey vx connectbot

Faheem Mitha

These small devices are quite the engineering feat. I wonder what kind of materials science goes into a touch interface like that.

@derobert thanks, I'll try it. If I can figure out how to get a wireless connection off my router.

yes, I have a wireless thing. But it's turned off. Sensibly so.

casey

@FaheemMitha just a capacative layer on top of the screen that can resolve multiple distinct touches.

the technology has had ~50 years to mature to this point

Faheem Mitha

I have a vague memory you can do encryption for wireless connections, but I don't see an encryption option here.

@casey I see. I haven't really run into this before now. Regular computers don't bother with it.

ok, I see the encryption stuff.

casey

19:23

@FaheemMitha go laptop shopping, that used to be true but not as much these days

at least among the win 8 laptops, touch is a fairly common feature

Faheem Mitha

Actually, I was reading about how SIMs connect to the network. Looks like the use a form of pgp or at any rate public key encryption.

@casey Oh, how recent is this?

casey

@FaheemMitha a year or two? not sure

Faheem Mitha

I don't really keep in touch with the latest gadgetry. Maybe if I travelled more.

@casey Ah, fairly recent then.

casey

mass market was probably about the time win 8 came out

Faheem Mitha

@casey ok

I wonder if this capacative stuff was also funded by the DOD.

They seem to behind a lot of computer related stuff. Maybe no 100%, but I'd say upwards of 80% of fundamental engineering research related to computers.

I guess the touch alters the local electrical properties of the material. Which means it is probably really easy to fry one of these gadgets.

This little screen here does seem remarkably sensitive, but this might be like 5th gen technology at this point.

casey

19:28

Capacitive sensing

In electrical engineering, capacitive sensing is a technology, based on capacitive coupling, that takes human body capacitance as input. Capacitive sensors detect anything that is conductive or has a dielectric different from that of air. Many types of sensors use capacitive sensing, including sensors to detect and measure proximity, position or displacement, humidity, fluid level, and acceleration. Human interface devices based on capacitive sensing, such as trackpads, can replace the computer mouse. Digital audio players, mobile phones, and tablet computers use capacitive sensing touchscreens...

derobert

@FaheemMitha there should be one. Hopefully WPA2-AES.

Faheem Mitha

@derobert Yes, I found something.

casey

if the something you found was called WEP, then that is essentially nothing

Faheem Mitha

I'm getting a choice between AES and TKIP+AES.

derobert

AES

Faheem Mitha

19:30

That's the WPA/WAPI Encryption field.

Network auth is either WPA-personal or WEP.

derobert

TKIP has some flaws (and for some reason they're only giving the full name of TKIP+AES, presumably the other is CCMP+AES)

casey

WPA

Faheem Mitha

oh, and WPA-enterprise. Whatever that is.

derobert

WPA personal if you want to set up a password. WPA enterprise if you want your next task to be "configure RADIUS, PKI, etc."

Faheem Mitha

WPA enterprise is either WAP2 only or WPA only. Or auto.

casey

19:32

dont pick enterprise

derobert

WPA2

Faheem Mitha

@derobert ok.

derobert

@casey Maybe he wants to configure RADIUS!

casey

its more work than you could possibly want to do for fun

@derobert maybe :)

derobert

It is more secure... :-p

Faheem Mitha

19:32

WPA personal are the same options. Is the WAP2 a typo?

derobert

Probably. They surely mean WPA2

Faheem Mitha

@derobert ok.

derobert

WPA2 CCMP-AES is, AFAIK, actually a secure protocol.

Provided of course you use a sufficiently-long randomly-generated key.

Faheem Mitha

@derobert that's nice. Now time to set a passphrase.

I usually use sentences.

Sometimes also math statements. But sentences are easier.

derobert

Remember you will get to type this on the phone. Sentences (at least if they make sense) don't provide that much entropy per word... only a few bits.

Faheem Mitha

19:35

There are also beasts called resistive touchscreens.

casey

@FaheemMitha yes, but you don't want a phone with one of those

Faheem Mitha

@derobert Apparently I can hook a keyboard up to the thing. Haven't tried that yet.

@casey no?

derobert

@FaheemMitha Yes. They're not as nice as capacitive (less sensitive, less accurate, often much more pressure required), but they work in some situations where capacitive will not.

Faheem Mitha

@derobert ok

@derobert like gloves?

or a stylus.

derobert

@FaheemMitha Yep. Or with plain plastic styluses.

Faheem Mitha

19:36

@derobert ok

So are the people who work on this stuff materials scientists or electrical engineers?

casey

@FaheemMitha no. if you have ever used a pay kiosk at a store (the kind you swipe a credit card, hit "credit" and sign), note the poor touch response, especially if you use a finger instead of the stylus. those are typically resistive (though the newer ones I'm seeing are capacitive)

Faheem Mitha

@casey yes, those aren't very sensitive. you have to kind of lean on them.

casey

@FaheemMitha yes, you don't want that on a phone.

Faheem Mitha

@casey true

@derobert They do if the sentence is long enough.

derobert

@FaheemMitha Both. The glass was no doubt designed by materials scientists (etc.) to be scratch-resistant. The EEs figured out where your finger(s) are.

Faheem Mitha

19:39

The great thing about a phrase is that even a long one is easy to remember.

derobert

@FaheemMitha a few bits per word. So yeah, a long enough sentence can.

it's not something you'll be typing routinely, you don't need to remember it. Just write it down somewhere.

Faheem Mitha

@derobert Hmm. I figure the materials science people would also need to be involved in actual materials manufacture.

derobert

Probably. I'm sure there are a lot of different types of scientists/engineers required to make that display.

Faheem Mitha

@derobert Ok. Actually, I like sentences I can just remember. I remembered my gpg passphrase after 2 years. Without thinking about it in between.

It is a crushingly banal sentence, but probably nobody would ever guess it.

@derobert i was going to a Debian key signing. Remember me mentioning the release party?

derobert

not a question of guessing it, it's a question of how many sentences you have to try to brute force it

Faheem Mitha

19:42

I only got one signature offer, and even that was not complete. Bummer.

derobert

weird, was there not a keysigning there?

Faheem Mitha

There were two people I was hoping to meet that didn't turn up. However, I'm sufficiently impressed there is even a Debian scene here at all.

@derobert No, just me. Apparently.

derobert

/me wonders if its ok to set up a bot to VtC kali-linux. Ok, not really, I'm sure it's not, but I can dream...

2

Faheem Mitha

I went over there, and asked about key signing. The sole DD present said, sure, I'll sign your key. I emailed him his key, signed by me, and encrypted, but he has not responded.

derobert

@FaheemMitha ah, well, give him a few days.

Faheem Mitha

19:44

@derobert If the sentence is 20 odd characters, wouldn't it be difficult to brute force it?

My gpg sentence is 23 char I think. Probably better to have no spaces, and I don't.

@derobert I will.

@derobert How hard is it to write bots to manipulate web pages?

derobert

If it's made of words, its easy enough to grab a dictionary and start trying combinations of words. Even better if you can use grammar rules to try sensical things first.

Faheem Mitha

@derobert I think it would have to be an automated attack.

No human is likely to come up with the correct sentence by experimentation.

derobert

Yep. That's what you worry about. Anyone serious about attacking it will use computers to do so.

Faheem Mitha

@derobert sure.

If the NSA has reason to be seriously interested in me, then I'll worry.

derobert

And if you're targeted beyond automated attack, well, an xkcd cartoon about a wrench comes to mind.

Faheem Mitha

19:47

But I am not Julian Assange or Edward Snowden.

Or Glenn Greenwald, even.

@derobert Which one is that?

@Braiam likes to post xkcd here, so why not others? :-)

derobert

xkcd.com/538

hmmm... /me wonders why that failed to onebox

... that should at least

Faheem Mitha

@derobert :-) Not sure if I have seen that one.

@derobert well, if I want to invoke Randall to make my point for me, how about

To anyone who understands information theory and security and is in an infuriating argument with someone who does not (possibly involving mixed case), I sincerely apologize.

Dammit, where is my one-boxing?

Hmm, doesn't like https.

casey

@FaheemMitha yes, but I know how your passphrase is around 23 characters of common words with no spaces inbetween.

so I need not guess randomly, I can just try every combination from a big dictionary

Faheem Mitha

@casey Yes, but you're not the NSA. Are you?

/me is a little worried.

casey

no, but you've constrained the problem sufficiently that perhaps I don't need to be

Faheem Mitha

19:53

Is spaces between good or bad?

I figure good.

derobert

That's 4 random words, which he calculates as 44 bits of entropy. That's reasonable for online attacks... offline attacks (e.g., attacking the hash) is much faster. Instead of 1000 guesses per second max, tens of billions per second is reasonable.

casey

for example that xkcd is a lot of entropy but if I know it is just 4 dictonary words concatenated, that is a simpler problem

derobert

@casey that's the entropy he is using: you know it's 4 random words

Faheem Mitha

@casey true, but hopefully that information wouldn't come bundled with the passphrase.

@derobert Well, yes, Like I said, I'm assuming the NSA doesn't find me interesting.

Apparently they have computers. Lotsa computers. And they use them. And they don't care about the law or ethics or any of that pesky stuff the rest of us worry about.

derobert

44 bits of entropy / 4 words means he's counting 11 per word. So he's only using ~2000 common words

Faheem Mitha

19:56

One of my UNC classmates took a job at the NSA. He wasn't very nice.

Not fun to think those kind of people are in charge of important decisions.

derobert

@FaheemMitha That's not the NSA. If the NSA was your adversary, you should assume hundreds of trillions/second, at least.

Probably a fair bit more.

Faheem Mitha

@derobert Um, ok.

derobert

Tens of billions is spending a few bucks with Amazon...

Faheem Mitha

hundreds of trillions/second seems a little excessive.

Do they do massively parallel computer cracking?

derobert

No one knows what NSA does. But getting into the billions of hashes/sec range is well under $1000 now.

Faheem Mitha

19:58

But realistically, if they are trying to crack a connection remotely, they don't have the actual encrypted thing in their hands, do they?

casey

just ask the bitcoin miners :)

@FaheemMitha anyone driving by can capture all your encrypted traffic and save it for analysis

Faheem Mitha

Which they would need to do to do very fast guesses.

@casey Well, Ok, but they would not be trying to attach the encryption directly.

casey

thats why WEP is terrible to use, they don't need to save anything and can crack the password without even slowing down as they drive by

Gilles

@derobert only for really crappy password hashing or a very powerful adversary. A decent password hash should take about 0.1s/processor/attempt.

Faheem Mitha

Though I don't know anything about crypto. I wonder if people discuss this kind of thing on crypto.sx.

Gilles

20:00

@FaheemMitha lots of stuff on the topic on Information Security

Cryptography tends to be more advanced than that

Faheem Mitha

@Gilles Ok.

derobert

@Gilles well, we're talking WPA2 here. Which uses 256 rounds of PBKDF-2, if I remember correctly.

David Freitag

@derobert Something similar to this has been done before. erwaysoftware.com/charcoal automatically parses comments and based on certain keywords it flags them.

Faheem Mitha

@Gilles more advanced than what?

@derobert and @casey you seem to know quite a lot about this stuff.

Gilles

@FaheemMitha it's mostly about crypto primitives, not about how to use them to hash passwords

David Freitag

20:02

In case you are wondering, it's on github. github.com/Charcoal-SE/Charcoal

Faheem Mitha

Is this the sort of thing they teach in US high schools these days?

Gilles

@derobert 256 rounds is too small, but it would still take a powerful adversary to reach 1e10/s

Faheem Mitha

@Gilles ok

casey

@FaheemMitha not back when I was in high school

Faheem Mitha

@casey ok

casey

20:03

but then again I taught the CS class I was taking my senior year :)

crypto is math though, not CS

Gilles

@casey more CS than math, but it does involve math

Faheem Mitha

@casey Depends on your defn of math. I know some math people who would disagree.

David Freitag

Purely theoretical crypto is pure math, applied crypto is both.

casey

@Gilles yea, I shouldn't discount the CS as much. The implementation is only as strong as the code

derobert

@Gilles Well, single GPU SHA-1 performance is over 5 billion/second now, quickly looking it up. So I'm not sure what that'd cost on AWS, but yeah, it'd cost some money

Faheem Mitha

20:05

Though I hear they use elliptic curves in crypto these days. And computational algebraic geometry and number theory, more generally.

@derobert what kind of GPU?

@DavidFreitag True. Like Daniel Bernstein.

Though I've no idea what he actually does.

derobert

@FaheemMitha Radeon R9 295X2 according to a table I found.

Gilles

@derobert With gvfs and udisks, do you know if /etc/fstab is parsed, or if it's the mount options that count?

Faheem Mitha

@derobert oh. is that expensive?

derobert

@Gilles I think it's /etc/fstab

Gilles

I don't understand where it looks at the volume and where it looks at the mount point

derobert

20:06

@FaheemMitha ~$650 or so.

Faheem Mitha

@derobert that's not much.

David Freitag

an R290X can achieve 27333.8 MH/s

gist: World's First 8x R9 290X oclHashcat Benchmark, 2013-12-29 14:38:38Z

8x290X

Gilles

And do I have to restart something (udisks?) or can I test by updating /etc/fstab and then opening the file dialog?

Faheem Mitha

@DavidFreitag does everyone know more about this stuff than I do?

derobert

@Gilles Not sure if you have to restart udisks. You don't have to restart anything (or even reopen the file dialog) if you do it the udev route. (Well, you have to udevadmin trigger, of course)

@DavidFreitag Apparently I need a more up-to-date table.

@DavidFreitag wait, that's 8x of them

David Freitag

20:10

@FaheemMitha I would say it's a coincidence. I happen to know a few people who build multi-GPU rigs for various crptyo-related projects.

Faheem Mitha

@DavidFreitag oh. I guess crypto is getting to be big business these days.

David Freitag

@derobert Right, I forgot that. It breaks down each GPU though.

Gilles

Hmmm

I just tried something I should have tried yesterday

Different users have different volumes listed

Faheem Mitha

20:37

It's a bit unfortunate that Mohsen now has edit privs on U&L.

shirish

20:52

Faheem, were you able to go to the Debian release party ?

@FaheemMitha ^

Gilles

*($^()*$#)($#

shirish

also does anybody want to see if the tag-addition and tag-edit are good or not unix.stackexchange.com/tags/apt-file/info

filed about apt-file as it is a necessary tool in a sys-admin toolkit but apparently not many people know about the tool .

Faheem Mitha

@shirish I did. Wasn't much of a party. You didn't make it, nor did the IITB guy. Forget his name.

Kumar.

shirish

were you talking about Kumar Appiah

Faheem Mitha

Kartik said he would sign my key, so I signed his key and emailed it to him using cass, as #debian-mentors recommended. Though he told me to upload to the servers.

But #debian-mentors made disapproving noises about it, so I used cass instead.

shirish

20:59

ah yes. I am guessing because people who are in IITB rarely feel the need to venture outside as it is very nice green environment and just next to the lake.

Transcript for