« first day (1509 days earlier)      last day (3804 days later) » 

17:00
Stallman's more like free software than open source.
and if it does it doesn't necessarily mean a bad one
ok but ^
the Open Source Initiative is a specific organization with specific goals
Stallman has explicitly espoused that his beliefs are not shared by the OSI and vice versa
are there any reasonably libre oriented open source movements besides stallman's?
17:01
lol
ignore the negative connotation of Nazism; all I'm saying is that you can't say "Stallman-style Open Source" any more than you can say "Franklin D. Roosevelt-style Nazism" because you are using a person's name immediately next to something that they have spent their entire life trying to distance themselves from
don't use peoples' names so disrespectfully
his libre software is still open source, whether he likes it or not, it's not closed source
it's just open source ++
Right, but you have to respect the person's wishes for what they want their name to be associated with -- a person's name and reputation is like a company's branding; you have to respect it
yeah but he's not selling open source.
you can say "Copyleft-style open source", and that's fine, because Copyleft is not a trademark or the name of a person or organization; it's a concept
and that's fine
17:03
@jokerdino the term selling is a bit loose. He isn't selling anything in the typical sense of the term selling
but to associate Stallman's name with open source is just !!no
Libre style open source
@barlop i don't mean selling in the literal sense.
that's fine, too
copyleft is a bit of a headache ;-)
people have enough trouble understanding copyright!
i'm pleased I found a title for what i'm referring to that i'm happy with and is not disrespectful to stallman
17:05
really within the open source camp, the sub-groups consist of: copyleft-positive (you should use copyleft licenses!); copyleft-negative (you shouldn't use copyleft licenses!); and copyleft-neutral (IDGAF about copyleft)
the FSF falls into the former; the BSD camp is copyleft-negative; and Apache is kind of copyleft-neutral (as is the WTFPL and friends)
Linus Torvalds is copyleft-positive, but disagrees with the added baggage related to patents in the GPLv3
after looking into that one I agree with stallman over torvalds though it's not at the top of my head..
oh yeah.. you can't change software because they've locked the hardware
Tivoization
and it's not like a CD-R where it's a limitation of the hardware.. but it's an intentional lock if I recall
there's another axis on which "Open Source" people can disagree :P
I heard Torvalds make an argument that it's like with a CD-R
and he mentioned those that are against tivoization.. say that if one person can write then everybody should.. and that that's an insane argument..
17:10
there's Tivoization-positive Copyleft-positive (use copyleft, but tivoization is fine); Tivo-negative copyleft-negative (don't use copyleft, and tivoization sucks); tivo-positive copyleft-negative (tivoization is fine, and don't use copyleft); and tivo-negative copyleft-positive (tivoization sucks, and use copyleft) :P
But a better argument.. would be that it's an intentional lock thus wrong i'd go with that.. stallman camp over torvalds on that
well, I'm tivo-negative copyleft-neutral, but I disagree with open source projects that don't have some kind of policy or license or something that will go to some length to prevent the project from being proprietarized in the future
"it's open source for now" or "this little bit is open source but the actual good stuff is proprietary" are dick moves
copyleft makes those impossible but you can also make those difficult/impossible using other techniques
morning /ugt
can someone please send me in the right direction for joining/linking wlan0 + eth0 (i get internet on eth0, and want to add to the speed by connecting to some hotspot) maybe browsing on eth0 and downloading on wlan ?
bridging network connections?
yes!
Bob
Bob
17:16
!!tell 17815122 google linux multiple internet connections
1
A: With two interfaces, how can I make one the default route and the other an alternate?

Jan MarekFirefox can go through wlan0 only in this cases: 1) If it would like to use address 169.254.x.x/16, which is automatically added link-local address - you can disable it by uninstall avahi* stuff. 2) If you would like to go to some address in the 192.168.25.0/24 network. If you would like to us...

1
Q: Linux split access (multiple internet connections w/ load balancing)

MikeShort story: I've got a faulty cisco 1811 on my hands. It's a dual wan fixed configuration router. I had it setup with route maps to do load balancing. It works but every couple of hours it shits itself and needs rebooting. I need to take it out of it's current position on the network. So what I...

bridging the eth0 with the wlan0 won't do what you expect
Bob
Bob
o.O
I... only gave the command once.
@allquixotic Cavil seems to be skipping.
!!info
17:21
@allquixotic I awoke on Fri, 19 Sep 2014 14:57:49 GMT (that's about 4 days ago), got invoked 16 times, learned 88 commands
bug with !!google maybe
Bob
Bob
@allquixotic Strange delay, too.
!!google fish filenotfoundexception
I'm wondering if the words are being parsed as two arguments and replied to separately
Bob
Bob
17:24
@allquixotic Nup.
that's Canadoogle anyway (Canadian Google) lol
I get .ca domains all the time at work because I'm stunneled in
must send Google for a trip when I sign in with my google account on the dedi =D
"Welcome, user! *performs IPGeo* ... from Canada!"
Galaxy S5, every 2 minutes seconds: "User is in Baltimore, MD; tracking will continue"
Google: "????? :("
"But there isn't a Maryland in Canada!"
I have given environment details. I would prefer using Visual Studio as we are using it from long time and it had worked fine. — Deep 3 hours ago
> I am using Windows Server 2003 SP2 and "Microsoft Visual C++ 2003". The same environment I am able to build Apache 2.2.22 without any issue.
> 2003
gcc was at version 2.95? or so in 2003... and nobody expects that to work with modern programs
@Bob update on the gimped 6 GB / Xeon E5520 dedi we're trying to stand up @ work: they're ordering 8x8GB DDR3 UDIMMs for it yaaaay \O/
Bob
Bob
lol
CPU's still worse than mine! :P
it has 16 slots so the guy was like "we'll put in for 8x8 and see if it's enough; if there's too much memory contention we'll order another pair and fill it out"
upgradeable 2009 low-end server tech ftw
looking like the host will be ESXi
we need both Windows Server and RHEL
and I think they want me to be the sysadmin :D
add that to my laundry list of duties
no KVM-o-IP, so I'll have to freeze in our datacenter closet on the physical console til I get RDP up
well, RDP and SSH
:17815676 ROFL
who knows, I might share terminal with you ssh'ed into it over gateone ;p
if a bitcoin miner accidentally managed to make its way onto the server....
Bob
Bob
17:44
I do hope you aren't typing that at work :P
what harm would it do?
@Bob I'm only kidding, anyway
but I am :S
Bob
Bob
@allquixotic Yea, but some people won't know that.
@allquixotic Probably not much good, either. Mining on an old, low-end server CPU?
Well, at least it's not MIPS.
@Bob if they're reading this chat they'll know from the fact that I just typed it so they'll have no excuse not to know it
@Bob hehe :P yeah I don't think it has a GPU
I have professional integrity! I won't do anything stupid
I do stupid things to my own hardware
not to someone else's
oh YEAY!!!
they're upgrading our mboxes from 100 MB to 2 GB ;D
emailfs is alive, baby!
(most coworkers use email as a filesystem, hence the term)
and no, I don't mean this, whatever the crap that is (with editor working files, to boot)
ROFL
> emailfs-driver.ko
not touching that with a 10-mile pole with a VM hanging off it
Bob
Bob
@allquixotic Now I'm picturing someone swinging an extremely long pole with an old greenscreen terminal hanging off it.
@Bob :D ... I have to admit I was pretty thrilled with the concept of shreding an email until I looked into the code of that driver, though
"hate spam? don't just rm it; shred it!"
he didn't even try to use fuse :(
 
1 hour later…
19:22
Newest Intel HEDT i7-5960X has 12 cores on die, but only 8 cores are enabled
Intel is taking a server-grade Xeon die and scaling it back for consumer use
The base clock rate is 3.0 GHz
The idea is to enable higher clock rates
Enabling all 12 cores would dramatically limit maximum clock speeds and further increase the cost of the processor
The result would actually be worse performance for most consumer applications
The Xeon E5-2658 v3, built on the same die, runs at 2.3 GHz with all 12 cores enabled—this is as good as it gets for this die
Bob
Bob
O_O
Actually, 2.2 GHz, not 2.3
To drive the point home, Tom's Hardware shows how the 6-core parts often outperform the 8-core part in gaming applications
19:51
@DragonLord I'm sure OCers who use nitrogen / liquid cooling would love a dual-core or quad-core chip with the same TDP as one of those 12-cores, but with really, really fat individual cores for maximum single-thread perf...
5 GHz :D
@allquixotic That's why the i7-875K/2700K/3770K/4770K/4790K (the Intel mainstream performance platform) has always been so popular.
Bob
Bob
@DragonLord Oh, and the whole not-costing-a-small-fortune part probably helps.
@Bob The Intel HEDT parts (i7-990X/995X/3960X/3970X/4960X/5960X) are produced from much larger Xeon dies, which alone dramatically increases manufacturing cost
Larger die = more room for manufacturing defects = higher cost
Often exponentially as the die area increases
Bob
Bob
@DragonLord + fewer chips per wafer/production run
It's more than just fewer chips per wafer.
20:04
meanwhile, IBM actually makes the world's fastest microprocessors, even if they're on a fab dating back to Intel's Sandy Bridge era
The zEC12 microprocessor (zEnterprise EC12 or just z12) is a chip made by IBM for their zEnterprise EC12 mainframe computers, announced on August 28, 2012. Manufactured at IBM's East Fishkill, New York fabrication plant, the processor began shipping in the fall of 2012. IBM stated that it is the world's fastest microprocessor and is about 25% faster than its predecessor the z196. == Description == The chip measures 597.24 mm2 and consists of 2.75 billion transistors fabricated in IBM's 32 nm CMOS silicon on insulator fabrication process, supporting speeds of 5.5 GHz, the highest clock speed CPU...
Bob
Bob
@DragonLord Which is why I said +.
This is the same reason DSLRs with full-frame sensors cost far more than those with APS-C sensors. Larger dies not only mean fewer chips per wafer, it also means defects are more likely to occur on each die
> up to 90 instructions can be "in flight"
Bob
Bob
@allquixotic I wonder how the actual performance compares.
@Bob umm.... the zEC12 will be faster? WAY faster :)
Bob
Bob
20:06
@allquixotic But how much?
@allquixotic ...not necessarily a good thing. See: NetBurst :P
The NetBurst-based Pentium 4 was notoriously inefficient
@Bob well, when you have 5.5 GHz of clock cycles, you can afford to throw a few away :P
Bob
Bob
@allquixotic That's what they thought with NetBurst pushing 3-4 GHz (a lot for its time).
A branch prediction error means the whole pipeline gets tossed out—wasting as many as 31 clock cycles
@Bob NetBurst didn't have hilarious amounts of cache, either, and it spent a lot of its time waiting on memory accesses
Bob
Bob
20:08
I mean, I'm not saying they definitely suffer from the same problems. Just that it's not necessarily a good thing.
I'd really be interested to see some actual benchmarks, though.
don't discount NetBurst's architecture, either; although the Core architecture was a major departure from NetBurst, Nehalem revived major design philosophies based on NetBurst
it wasn't that NetBurst was bad, just that the implementation specifically chosen on an anemically low-cache uniprocessor with slow RAM was pretty foolish (and they were bad at controlling things like leakage and heat dissipation back then; we're better at that now)
Bob
Bob
Can't find any :\
Probably too expensive for anything interested to even think about getting (access to) one.
apparently Tim Cook is getting some z/OS mainframes :)
Bob
Bob
wat
(TLDR: Apple is going to put z/OS muscle behind some of its cloud services)
well. it's not completely a surprise: Apple has been in love with IBM before
remember the Power architecture on Macs?
Bob
Bob
20:13
> Apple’s legendary ... developer platform
glances at Obj-C
actually I would really like to see IBM stand up to Intel in the desktop and dedicated server markets
imagine OVH or Hetzner or Softlayer offering zArchitecture dedis for $200/mo running CentOS or Ubuntu
hexacore :D
Intel needs competition
they're lazing around atm
@allquixotic The most likely competition is from ARM, not zArchitecture
@DragonLord I definitely think IBM can compete, but I doubt they want to -- and there's also the matter that they probably can't get the costs down to levels affordable by people who buy 1U servers and ATX towers
Bob
Bob
Eh, we've got both ends.
AMD has 8-core ARM Cortex-A57 parts for energy-efficient servers
Bob
Bob
20:17
Cheap, low-power ARM. Lacks in performance.
Expensive, power-hungry IBM. Lacks in price and power consumption.
x86 is pretty much the perfect combination right now, for desktop and laptop computing.
and Intel, which is "just right" in price/performance and okay on power
@Bob Don't underestimate the performance per watt of an ARM chip.
Bob
Bob
IBM's probably in a better position to bring it down.
ARM's struggling to improve their performance.
@DragonLord Hm? Don't overestimate it either.
Intel's focusing more on the low-power end now.
ARM has less motivation to improve performance than to reduce power consumption -- almost everyone who's doing anything with ARM is obsessing about how to get performance as if you had a modern Snapdragon 801 running full-bore, while sipping battery life as if it were basically idle
Bob
Bob
IBM's got basically free reign over the high-power end, and they could compete with Intel at the general purpose server level (Xeons).
20:20
they aren't trying to make ARM chips as fast as Core i7s; they're trying to make ARM chips as fast as today's ARM chips that use half the power (or less)
Bob
Bob
Probably not on consumer devices, though. Not without drastic changes.
(though they are still trying to improve GPU performance a lot)
@allquixotic Hence, NVIDIA pulling tricks like an invisible, power-saving fifth core on low-leakage silicon in their Tegra SoCs
They call it "4-Plus-1"—completely invisible to the operating system, and used in low-demand situations to extend battery life
In Tegra 4 and Tegra K1, it's basically a low-clocked Cortex-A15 on special ultra-low-leakage silicon limited to less than 1 GHz for maximum efficiency
they need to do something about background services eating up all the resources (causing intensive GCs or whatever) which leads to bluetooth dropouts... still happening on my S5 even after their apparent patch to fix bluetooth dropouts
Bob
Bob
@DragonLord Apparently that one (A1100) has a TDP of 20W.
Quite high for an ARM part, quite low for AMD :P
I wonder how it actually performs. And how much it costs.
It'd be competing with i3s and i5s (heck, ULV i7s) in that TDP bracket, though with the ability to address far more RAM.
Oooh, didn't know Intel's still doing server Atoms.
It'd be more directly competing with the Atom C2758, also 20W TDP.
The chip itself might be about half the price.
The A1100 can address 128 GB; the Atom 64 GB of RAM.
Ouch. CPU performance, it's going to be competing more-or-less with an ULV Haswell i5 (multithreaded).
Guess it's wait and see.
Hm. The A1100 has 8 MB of L3 cache. I wonder how much of an advantage that would be (Atom has none). Same L2 size, dunno about L1.
20:38
It depends on the size of the program you run
For RC5 like tasks: Meh
For large matrix multiplications (e.g on HPC): Quite a lot if they fit in the cahce.
Not as much if they do not fit
Bob
Bob
Hm. The A1100 is not out yet. The Atom I was comparing it to has been out for about a year now.
If AMD don't get it out soon, the next Atom will probably overshadow it.
god damnit... I hate that there are hundreds of little system incompatibilities in Solaris that just aren't there on Linux or even BSD
@allquixotic Back to the faster-but-fewer-cores discussions, if you want a dual-core processor that responds very well to overclocking, get the Pentium G3258
try to build llvm, fails because of some wacko binutils problem
Bob
Bob
It's also going to have to be either quite cheap or significantly better in some way to disrupt x86.
20:41
It's the only recent Pentium part to be fully unlocked for overclocking, to commemorate 20 years of the Pentium brand
try to build binutils, another weird error
Bob
Bob
I mean... given a choice between identical-in-every-other-way ARM and x86 for a server, most people will choose the latter.
Fits in the same LGA1150 socket as other Haswell parts, can hit 4.5 GHz on the stock cooler
The retail packaging for this processor bears the words "Unlock the Power Within", very unusual for a Pentium
I think that's what Jimmy Hoffa had
20:59
@allquixotic huh?
Bob
Bob
@allquixotic Didn't some guy have major issues getting xrdp working with lxde?
...I just installed lxde, updated my .xsession, and it works like a charm.
@Bob Dunno
might've been me
Bob
Bob
@allquixotic urk. it did break something, though
it installed network-manager, which completely fucked up my networking
21:21
@Bob that's the default behavior of network-manager :D
Bob
Bob
@allquixotic apt-get autoremove --purge network-manager ahhh all better :)
22:12
Guys I made a great realization! Perl is awful.
2
@nhinkle congrats on your Use of uninitialized value in method with known name at nhinkle.pl line 1
Bob
Bob
@nhinkle Better than PHP.
@Bob Perl Hyped-up on Pseudoephedrine?
less snot; more hallucination
22:30
@Bob Well, it is PHP
So, a goto is fine
Bob
Bob
@OliverSalzburg That's not really the fail here, though. The fail is that they even bothered optimising that.
As someone further down the thread said:
> If someone has difficulties seeing why optimizing a single jump in a loop is totally unimportant when your database/network/disk is down, has either some problems with BigO notation, or with priorities, and should not be allowed near keyboard without supervision.
Welp. Someone just satisfied Godwin's Law.
> i'm sure you'd like to tattoo serial numbers on us all and line us up for free showers too, while you're making the rules about who is allowed to do what
 
1 hour later…
23:52
0_0
@Bob The avatons are pretty nice ;p

« first day (1509 days earlier)      last day (3804 days later) »