/dev/chat

Michael Mrozek

18:01

@slm I don't know what kind of edits you guys are planning; you shouldn't change an old question to match a new question just so you can close it as a duplicate. If you want to slightly change the old question without changing what it's asking for, that's fine

derobert

@Kiwy I don't think there is a real reason (well, at least not with modern computers). It's just the way Unix is.

Commands are traditionally case sensitive, and almost always in lowercase.

Kiwy

@derobert I think it comes from the file system

because you often have file as argument so case sensistive

and it make more sense to be case sensitive with every argument

derobert

in C, you change strcmp(arg, "option") to strcasecmp(arg, "option") ...

given, gits subcommands were, I think, once actually different executables, so filesystem case sensitivity mattered there.

Kiwy

I wasn't sure is it the case for apt-get ?

derobert

But really its just tradition. Unix has always been case sensitive. So long-time Unix users expect case sensitive. I'd be surprised if I couldn't set a git alias for Status that did something different than STATUS and different than status

@Kiwy There probably is no reason for apt-get to be case sensitive in package names.

Aptitude, for example, does not appear to be

Actually, neither is apt-get

apt-get remove BASH <--- see if you can break your system!

but apt-get REMOVE doesn't work. Probably just because no one ever thought of it. I mean, first time I've ever considered it...

Kiwy

18:14

Well maybe the OP should be satisfied with my answer then :D

terdon

@derobert spoilsport... :P

Hmm, dunno why that appears as a reply to your post, I did not enter it as one.

derobert

@terdon when you use a plain @-ping, it shows as a reply to the most recent post by me. Don't know of any way to avoid that.

Well, that apparently did, there were two @ signs in there...

terdon

Oh well, thanks for the clarification on my A anyway.

Huh, let me test that @terdon

Nope sorry, I'll ping you @derobert @-2

Cool that works

derobert

Yep. And it still pings.

terdon

Cool, sorry for spamming ya

goldilocks

18:29

@MichaelMrozek I was thinking of changing the title to make it more general ("rotating files" instead of "rotating logs") and get rid of some of the extraneous details ("I have this app that does such and such blah blah blah") to make it more canonical.

terdon

18:45

@goldilocks why did you get so worked up about the close votes? All your arguments about not closing (posting in chat etc) are equally applicable to reopening.

Personally, I want Qs to be closed quickly and reopened equally quickly as needed.

As tends to be the case, why not just post here and ask us to vote to reopen?

goldilocks

@terdon That one's a bit of a mistake; I filtered slm via Hauke who made a very valid point about logrotate, and since the other question was about logs and smothered in remarks about logrotate I presumed that's what it was about. WRT the fact that questions in general can always be re-opened, if you miss an exit on the highway you can always turn around and go back, but it makes more sense and wastes less time to just get off in the right place to start with.

terdon

I was talking about your meta Q

I am in favor of quick to close, quick to open. I just didn't want to copy my answer from the linked one. Yes, in this case, as you pointed out, the question was answerable but it was not very clear. It would have benefited by a mention of the OS, a clarification of when exactly the error occurs, if it happens with all files or just one etc.

I find the best way to get an OP to clarify is to close and then reopen. Since any edit automatically pushes a post into the reopen queue it seems like a good method.

goldilocks

@terdon Well, I think I pretty much spelled it out there. If people are voting that fast to close a question that obviously should not be closed, it implies to me they don't really care and just go through the queue figuring if someone else said something should be closed, then we might as well close it. That was a new question, but without the Close queue I very much doubt it would have had more than one vote....

...in other words, the only reason those people looked at it was because it was in the queue. They obviously were neither interested in, nor, I think, understood it. However, you not understanding something does not mean it is unclear.

terdon

19:00

Perhaps. I would also have voted to close as unclear though and still stand by it. I did not cause I hadn't seen it but just because a question is clear to someone who knows the answer does not mean it is actually a clear question.

That's my point, I won't close as unclear something I don't understand if it looks like there's enough info there for an expert to understand it.

Yes, that judgment call may be wrong, but the alternative is to leave Qs open for ages because we'll never remember to go and revisit them.

goldilocks

Again, the reason that one set me off is I don't see how it could have been much clearer than it was to start with. All the information was there, and the context was also very obvious. What was really wrong with it was it was brief, and unformatted.

terdon

That's the whole reason for the new system of "on hold" and auto-reopen queue.

> It would have benefited by a mention of the OS, a clarification of when exactly the error occurs, if it happens with all files or just one etc.

goldilocks

@terdon Okay, but this also sends a message to posters that we'd rather toy around with interface features that address them directly. Like I said, I think the first "hold" vote should require a mandatory comment of explanation.

terdon

Anyway, as a general rule, I tend to agree with the closures, and that's the whole point of the queue, you get to vote to leave it open and you also get to vote to reopen.

@goldilocks not mandatory but I tend to leave one, yes.

goldilocks

@terdon Right, and the point of a highway is you drive on it. But the fact that you can get off anywhere, cross a bridge, and go back to a previous point does not mean that is the efficient or correct way to get from A to B.

I do not really see the "on hold" and "reopen" option as an excuse for closing questions at the drop of a hat etc. Rather, I see it as a means of correcting mistakes. The fact you have a backspace key, etc. should not encourage you to just type blindly all time. It allows you to make corrections, but it shouldn't be an excuse for making mistakes.

rather toy around with interface features that address them directly -> rather toy around THAN address them

terdon

19:12

Fair enough, I just did not find that closure wrong. I still thing the original question was unclear. And in any case, I feel that most closures are correct. We tend to close things that should be closed and that is just a message to the OP to fix things. I just feel you're overreacting a tad.

goldilocks

@terdon E.g. I think Ben Voight's second comment on that question is some really obnoxious bullshit -- that it's not really a question *even though I pointed out the title is "a grammatically and semantically correct question" -- because "it wasn't the question the poster actually wanted and accepted an answer to" (malarky) or because "The accepted answer doesn't even address the question in the title" (more malarky)...

If the motive here is to defend the right to close things for completely absurd reasons and people are okay with that "because we can always reopen", then the potential for a hold->reopen cycle has had some very negative consequences.

terdon

@goldilocks Agreed, 100% percent. That was uncalled for,

But then so was YOUR SHOUTING so you each got one wrong :)

@goldilocks And no! Not at all, all I'm saying is that many users, myself included, did indeed find the question unclear. I just looked at the original version and stand by that. It only happened to be clearer to you because you knew about the issue at hand.

goldilocks

@terdon True. Like I said I need a closet I can lock myself into on occasion and hammer out my feelings in ALL CAPS.

terdon

Heh, yeah, I read that and share the sentiment actually :)

derobert

Wow, a +73 wrong answer on the hopeless site... superuser.com/a/733086/1735

goldilocks

19:22

@terdon That's fine but that's why I wrote the meta post. It's ridiculous to say all questions have to be perfectly clear to 100% of the people who use the site. When in doubt, ask someone else, wait, or let it slide. There's enough people here with enough knowledge -- if you are certain a question is unclear/nonsensical because you do understand the specific subject matter, fine.

If that person is not you, s/he will probably be along soon enough.

terdon

@goldilocks I agree with that, no argument there. I have often done so. All I'm saying is that I don't feel that it is a big issue here and even agree with that particular closure. Remember that questions should also help others and not only the experts of the field.

@derobert wrong?

derobert

Yes. A strong password, turning off WPS, and using WPA2-AES makes it uncrackable.

terdon

uncrackable? Is that even a thing?

goldilocks

@terdon It did need some tidying up, but OTOH, WRT posterity and helping others, the accepted answer to that question, together with the original title (which quotes the error message exactly, this is how I found the coreutils bug report) would have been enough.

derobert

@terdon Sorry, uncrackable in the context of the question, to "most enthusiasts"

Not to the NSA.

terdon

19:25

@goldilocks OK, it's the cleaning up I wanted. I feel that closing is a good way to do so.

> Network security is a big subject, and not something amenable to a Superuser question, but the basics are that security is built up in layers so that even if some are compromised not all are - also, any system can be penetrated given enough time, resources and knowledge, so security is actually not so much a question of "can it be hacked", but "how long will it take" to hack. WPA and a secure password protect against "Joe Average".

He's not saying that "Yes any idiot can crack it" the answer seems quite reasonable to me but I'm no expert.

goldilocks

@terdon Instead of cleaning it up??? I dunno, that seems pretty obnoxious. IMO.

derobert

AFAIK, there aren't any serious flaws in WPA2-CCMP-AES. So with a strong password (e.g., 22 random letters & numbers), its clear Joe Average can't do it. It's unclear if NSA can. Maybe, maybe not.

terdon

@goldilocks Well, I'd have no idea how to clean it up, otherwise I'd have done it myself. As it was, had I seen it in time, I'd have probably asked the OP for details and voted to close as unclear, yes.

@derobert Yes, that's pretty much what the answer is saying.

> WPA and a secure password protect against "Joe Average"

I think you might have just read the 1st sentence only :)

derobert

I read the first sentence, and I expect that to actually be accurate. You're basically telling me I should read the rest of the post as saying the 1st sentence was a lie.

goldilocks

Gotta agree w/ derobert vis, if the question is "Can some script kiddie with Kali linux break into my LAN even if I do the absolute most to prevent it?" the answer is almost certainly not, although there's always a chance. I could attempt brute force and be right on the first guess.

derobert

19:31

Q: "I heard from a trusted computer security expert that most enthusiastic users (even if they are not professionals) using only guides from Internet and specialized software (e.g. Kali Linux included tools) can break through your home router security."

A: "Without arguing the semantics, yes the statement is true."

Q also specified strong passwords

goldilocks

I could win the state lottery 3 times in a row. Time can effectively run backward according to the laws of physics. It's just very unlikely.

terdon

@derobert Yes, but he then goes on to explain in quite a bit of detail that in fact this is basically paranoia and that WPA protects you from Joe average. The answer is really quite good. It just points out that uncrackable just means "will take too long to crack"

derobert

Well, actually, if the key is long enough, brute forcing it isn't possible.

Not just takes too long.

goldilocks

Semantically it's correct, but personally I would not have started out by dismissing that as non-essential to the argument. But it is a good answer.

terdon

@derobert really? I thought it was more like "would take longer than the life of the universe" and stuff. But that will depend on computing power and all bets are off with quantum computers right?

derobert

19:36

Physics requires a certain amount of energy to flip a bit. It's pretty tiny. But it places an upper limit on the amount of computation that can be performed: we can not use more energy than exists in the entire observable universe.

terdon

Ah and that can be computed and will never be affected by processor speed. Gotcha.

Still, QC circumvents that. Or at least that's what I understood from an hour long drunkish conversation with a PhD in the field.

goldilocks

That still doesn't make it impossible. You're assuming the worst case scenario. A brute force attempt does not have to complete all possibilities in order to succeed. If the password were so long there was not enough energy to make even one guess, methinks it would be impossible for it to exist in the first place.

derobert

@goldilocks Sure. You can get lucky. But that's incredibly unlikely.

And it gets half as likely each additional bit of key.

goldilocks

Unless I used up 65% of the universe's energy on storing my password, in which case not even I would have enough left over to use it.

@derobert Yeah in real terms I totally agree and so does everybody else, or internet commerce, etc. would be impossible.

terdon

@goldilocks Dude! Stop destroying our universe! Get your own!

goldilocks

19:42

This is the second time today I've run across a Moxie Marlinspike reference. I wonder how many times he has collected that $30 and said, "Well, I told you it wasn't guaranteed..." I guess it is really for penetration testing.

derobert

I guess I'm annoyed mainly by the first sentence...

But I fear any edit I submitted would be rejected, as not respecting the author's intent

mikeserv

How do you remember it?

The .65 universe PW?

Wait... Is it 27.3?

goldilocks

@derobert OTOH, it is within the ability of a normal human adult to grasp what a power of two sequence is without too much trouble. Like, maybe they had to skip a few episodes of American Idol in order to reason it out for themselves. Might make for a better world. If that is too much, let them wander around paranoid because some guy on an internet forum said, "Without arguing semantics...".

derobert

I'd write my own answer, but the other already has so many upvotes it'll never be displaced, and it'd be a fair bit of work...

But really, unless your adversaries are three letter agencies, a strong password (20-odd random characters) with WPA2-CCMP-AES is perfectly good.

If I wanted to break into that network, I'd try smashing the window.

So, too, probably would the three letter agency. Well, they'd pick the lock.

goldilocks

@derobert Pretty sure until even the three letter agencies are bound by reality there. They would have to resort to other means.

derobert

19:51

@goldilocks Three letter agencies may know of flaws in the crypto algorithm. Or in the random number generator, etc.

Or in the implementation, etc.

goldilocks

@derobert I don't buy the "flaws in the algo" possibility because it presumes they are the only ones that smart, etc.

derobert

But really, they'd wait until you're not home, obtain physical access, and backdoor it that way.

@goldilocks Its reasonable to believe NSA, etc. know of flaws other people don't, because NSA employs a lot of people to look.

And they keep all of their results secret.

But that desire to keep them secret probably means even if the can, they won't—they'd save an AES vulnerability (for example) for something important.

slm

@terdon @goldilocks still stand by the close vote on that one as well. My vote is signaling to the OP and community that this is unclear to me. If the OP or others want to fix it then go for it. I'm more then happy to retract and reopen but it's 1 guys opinion that the Q is unclear to me.

goldilocks

@derobert Probably much closer to the real g-man style activity. The NSA may employ a lot of people, but that presumes it requires a lot of people or that an excess of them will make it easier. If I'm an expert on haystacks and I say it will take me and my team 3 days to either find the needle if it's there, and I don't find a needle because there isn't one, 100 more teams will not make it exist.

Ramesh

@slm, @terdon, For this question,unix.stackexchange.com/questions/121473/…

slm

19:57

@goldilocks This is probably where our disagreement lies. I'm not for letting an unclear Q "slide". If it's unclear it needs some love, my voting to close is the signal.

derobert

@goldilocks Sure. If there is actually no vulnerability, you can't find one. But if there is one, you can have different people trying different approaches. They also have a lot of computing power, money, etc. to throw at it.

Ramesh

How about this option as a suggestion for freshers?

goldilocks

Also, look at Snowden. If you were an NSA scientist who found a loophole, you would not have to expose yourself at all by leaking it because it is something someone who did not work for the NSA could also have found. So I doubt they have relied on that ever, actually.

Ramesh

For beginners, how about turning off the caps key using `setxkbmap -option ctrl:nocaps` so that all the commands will be typed in lower cases.
After finishing typing commands, they can turn on the Caps lock key by typing `setxkbmap -option`.
Even after turning off the caps key, if they need to key in upper case letters, they can still hold shift key and type the upper case letters if needed.

derobert

@goldilocks Sure. If someone wanted to leak it, they could. But that applies to all kinds of secrets. Yet look how long they kept those secrets, despite what appears to be ineptness...

And of course, assuming they know of flaws, they're probably also using them for causes that their employees believe in.

(Also, don't they have a huge catalog of flaws in implementations? They could use one of those; if it turns out a certain packet causes a buffer overflow in the router... Or if the router leaks info via timing attacks, or EM, or...)

goldilocks

20:02

@derobert They kept those secrets because, again, if they did get out, there would only be one possible explanation: a leak. Vis. what employees believe in, I think that is something employers would like to believe and prefer, but the more expertise required, the less they can control things.

derobert

Obviously leaks happen. But that doesn't mean everything will be leaked, or will be leaked quickly.

goldilocks

@derobert Yeah TBH I would trust the individual box + open source SSL, and not the network. If it's leaving your LAN anyway there's not much left to trust other than that.

derobert

Some of the Snowden stuff was done a while ago. It's been a decade since they backdoored Dual EC DRBG. It took a few years before someone even pointed out the possibility...

goldilocks

Basically they can either break TLS, or they can't. I think maybe they can via man in the middle stuff WRT to most people, but not everyone, e.g., could the NSA spy on the NSA this way? I bet not.

Not unless team A had infiltrated team B.

derobert

Presumably, NSA protects itself from attacks it knows about.

And if they use TLS, they probably only use an in-house CA

goldilocks

20:10

@derobert Begging the question that the attacks it knows about can be defended against. I'm inclined to think they could be, otherwise they'd suggest changes in protocols in order to make it possible in order to protect themselves.

Like, if they know of a vulnerability, they are not going to rely on "security by obscurity", presuming it's safe because "probably we're the only ones that know", lol.

derobert

@goldilocks Well, or NSA could just not put secret stuff on WiFi networks.

Which, hopefully they don't...

Though they do allow AES to protect top secret documents. So there is probably nothing wrong with AES.

goldilocks

@derobert True but that's basically the "we broke into your house" method. I don't doubt they do under limited circumstances.

@derobert If I had stuff that was really important I'd encrypt it before transmission too, so even the man-in-the-middle would be stuck w/ garbage.

derobert

You'd have to solve the key exchange problem.

goldilocks

@derobert Yeah, you could not transmit them ever.

derobert

The man in the middle works because they can pretend (by having a untruthful certificate issued) to be you.

If you already have keys set up, through some other secure method, then yes you can use those, and they can't MITM attack them.

Of course, then you wouldn't need to "ask" a CA if the remote side's key is right.

Anyway, time to get food, then go home and play Stick of Truth.

goldilocks

20:21

Have fun!

Faheem Mitha

21:17

@derobert that question we worked on yesterday - should that be closed? The user reinstalled, nobody will ever answer that question now.

@MichaelMrozek what is the site policy towards unreproducible questions?

I don't see this as a close option.

Michael Mrozek

@FaheemMitha It's under "off-topic"

terdon

@FaheemMitha Here:

Faheem Mitha

@MichaelMrozek ah

@terdon ok, thanks.

one sec, let me bring up that question

1

Q: I'm trying to install steam on debian jessie/sid but I'm getting unresolved dependencies

I'm following the debian wiki on installing steam but I keep getting dependency issues. Debian tells me to edit my sources.list to: deb http://http.debian.net/debian/ jessie main contrib non-free My sources.list looks like this: Debian GNU/Linux jessie/sid \n \l # deb http://ftp.uk.debian.o...

Already had two close votes - added mine. Now needs 2 more.

Myself and @derobert spent some time trying to help the poster fix his system yesterday. Usually such issues can be fixed but not this time. Finally the poster reinstalled.

So, I think a close is the correct thing to do here.

Two more close votes, anyone?

I wonder who deleted all those comments.

terdon

voted

@FaheemMitha Michael probably, only mods can delete other people's comments.

Faheem Mitha

@terdon Thanks. One more.

terdon

21:25

@FaheemMitha or anyone else, could you please vote to close this:

Faheem Mitha

@terdon ok

terdon

0

Q: Convenient way to check if system is using systemd or sysvinit in BASH?

I am defining common bash files which I want to use across different distributions. I need a way to check if system is using systemd or sysvinit (/etc/init.d/). I need this so I run appropriate command to start the service. What would be safe way to check for this? I currently check for existance...

linux scripting systemd sysvinit

As a duplicate of this:

0

Q: How to determine which init system is used?

Is there a way to determine which init system is being used in Linux? (such as sysvinit, upstart, systemd, etc.) I don't care if it can be programmatically determined, I just want to know how I can figure it out. I'm aware of Bash- detect init system, but that is about detecting the init system ...

linux systemd init upstart sysvinit

Everyone else has voted to close as a dupe of the bash init question but if you vote for the other, both will be linked in the closed message

Faheem Mitha

@terdon ok, but the poster doesn't seem to think it is a dupe

terdon

@FaheemMitha Nevertheless, it is because there simply isn't a good way to do this. That's why I wanted to get both Q&As we have in the message header.

We were talking about this the other day with slm and some of the others and there just sin't a good way to do it. All the answers have already been posted basically.

Faheem Mitha

@terdon ok, i tried to do what you wanted. did it do it right?

terdon

21:30

@FaheemMitha yup:

Faheem Mitha

@terdon great

You would like multiple choices to put for what the question is a dupe of?

terdon

In this case yes, because the bash one is very limited and slm's answer on the other one is more comprehensive.

@MichaelMrozek perhaps these two should be merged?

Faheem Mitha

The steam question still needs one more more close vote. Anyone?

terdon

0

Q: How to determine which init system is used?

Is there a way to determine which init system is being used in Linux? (such as sysvinit, upstart, systemd, etc.) I don't care if it can be programmatically determined, I just want to know how I can figure it out. I'm aware of Bash- detect init system, but that is about detecting the init system ...

linux systemd init upstart sysvinit

Faheem Mitha

@terdon ok

terdon

21:32

and

8

Q: Bash- detect init system

This may have more to do with detecting operating systems, but I specifically need the init system currently in use on the system. Fedora 15 now uses systemd, Ubuntu uses Upstart, while others use variations of System V. I have an application that I am writing to be a cross-platform daemon. Th...

shell-script init portability

Faheem Mitha

@derobert @strugee @casey can you vote to close, please? unix.stackexchange.com/questions/121384/…

mikeserv

The guy who "just wants to know how to find out" should try "Google."

strugee

21:54

@FaheemMitha just to make sure: we just closed because he reinstalled, right?

Faheem Mitha

22:27

@strugee right. we could not figure out the problem, so he reinstalled. so there is no point keeping it open, because it is no longer reproducible. as per my comment.

the close reason given isn't ideal, but I did add a comment at the bottom for other people who come this way.

Braiam

23:19

@terdon ok, why a duplicated of a duplicated?

terdon

@Braiam Because both of those dupes have useful information, so I want both linked when someone finds that Q.

The end of the dupe path is specific to bash and has some possible answers, the other has a very good answer by slm which is probably the best way to go but that is closed as dupe of the bash one, so, I thought it would be a good idea to have both there.

Braiam

you should instead ask for merge...

Faheem Mitha

@terdon a merge would be a good idea, probably

@MichaelMrozek ?

terdon

Yes, I suggested that to Michael above

terdon

23:34

In the meantime, double closing like this is a very good idea I think. That way, the dupe points to both useful answers.

Faheem Mitha

@terdon True.

Random pop quiz: anyone here familar with the name of Semmelweis? I mean, before Googling it. :-)

@terdon do you want to point unix.stackexchange.com/q/121683/4671 to your canonical or shall i?

Braiam

> Why Linux community does not set some standard for package management so that each and every software can be in .deb format?

-_-

terdon

@FaheemMitha you may as well. Not sure it will help but it is likely

Faheem Mitha

@Braiam are you quoting someone?

terdon

@Braiam Yeah, had to laugh with that one :)

Linux? Standards?

Braiam

23:43

@FaheemMitha I think so :P

btw @terdon anything worthwhile to waste my 19 close votes?

in AU

Faheem Mitha

@Braiam who?

Braiam

@FaheemMitha there askubuntu.com/q/439600/169736

Faheem Mitha

I don't disagree with the sentiment. I never much cared for rpm

terdon

@Braiam not off the top of my head, no

Braiam

love rmano answer D:

> Is like to ask "why I can buy cakes and flour, eggs, sugar too? It's so easier to just eat the cake, why someone sells all that other things too?"

terdon

23:45

@FaheemMitha Nobody disagrees, it'd be great if we could all decide on one, any one hey!

@Braiam Yes :)

Faheem Mitha

Oh, having read the question, the poster doesn't seem to understand the difference between binary and source.

he is asking why people use a tar.gz format

terdon

@FaheemMitha Or anything else. Fair enough, just a new arrival from Windows land

Faheem Mitha

@terdon I'm glad they are arriving there and not here.

Braiam

IMO the only distro that uses tar.gz is Arch/Gentoo

Faheem Mitha

@Braiam no, all the binary distributionis do too, as source format

@Braiam i don't know the answer to this, but what is gentoo's binary format, anyway? they don't use deb or rpm

Braiam

23:48

err debian uses a boilerplate of .orig.tar.(xz|gz|bz?) .dsc and .changes

@slm do you know what the heck did CentOS here? askubuntu.com/questions/439519/…

Faheem Mitha

@Braiam right. my point exactly

@terdon i think wrt that grub question, it really makes sense to transplant your answer to a suitably written question, and use that as the canonical. per that posters reaction.

the same goes for other examples, of course

Braiam

he's asking the wrong question :/ askubuntu.com/review/close/239913

can anyone search for me a shortcuts/.desktop Q on AU?

Transcript for