/dev/chat

5:36 AM

@Gilles If you could please let Java, many JS implementations, etc. know this, it'd be appreciated. Especially Firefox's JS.

I realize that the CS folks have figured out how to do GC with no perceptible pauses... unfortunately, far to many of the real-world implementations haven't heard yet.

(OK, openjdk has mostly solved the pauses, even if you abuse the GC pretty hard. But it wasn't even 10 years ago the JVM required careful tuning to keep them under control, depending on the application.)

And really, bad GC implementations are annoying because good garbage collection is awesome.

Faheem Mitha

5:58 AM

@Fabby How does a HVAC differ from a split AC standalone unit?

@Gilles @derobert what do you think of SBCLs garbage collection?

derobert

Haven't used it.

Faheem Mitha

@derobert haven't used SBCL?

derobert

Nope.

CC(T+2130256.8) max pause: 1066ms, total time: 1568ms, slices: 47, suspected: 12013, visited: 468679 RCed and 1173584 GCed, collected: 9411 RCed and 673 GCed (673|0|97 waiting for GC)
ForgetSkippable 16 times before CC, min: 1 ms, max: 65 ms, avg: 16 ms, total: 261 ms, max sync: 54 ms, removed: 22413

@Gilles if you could tell my Firefox that kind of terrible performance hasn't been seen since the 60s, it'd be appreciated.

Actually, that's good as far as it goes—normally, it's the same amount of time to collect 0.

Faheem Mitha

@derobert I didn't know JS uses garbage collection. What about perl?

derobert

Over and over and over and over... Every few seconds. I seriously am tempted to recompile Iceweasel w/o the cycle collector, and see if swap is faster. Probably.

@FaheemMitha perl to, though perl5 is mostly reference-counted

Faheem Mitha

6:02 AM

@derobert oh

derobert

which isn't so great performance-wise overall (reference counting turns out to be expensive), but predictable... And of course, it's a script language, so a few percent performance hit isn't a huge issue.

Faheem Mitha

I've had problems with the SBCL GC. And the developers give the impression that they don't really understand how it works. So, yes, I can see there could be issues.

One SBCL bug I filed was closed with essentially the comment: "I sorta fiddled with stuff and it went away". Not literally, of course.

i386 seems to be more problematic. Something to do with registers.

derobert

@FaheemMitha Fun :-/

Faheem Mitha

@derobert indeed.

I also had some difficulty even getting them to acknowledge there was a problem.

CL is a great language. The implementations leave something to be desired.

derobert

I should try filing a Firefox bug report. I asked it to save a concise log, its 2.3 gigs. I bet I could take out Bugzilla trying to add that!

Faheem Mitha

6:31 AM

@derobert :-)

@derobert here is the SBCL report I had in mind, I think.

(gc :full t) has no effect under certain circumstances

Hi, $ sbcl --version SBCL 1.0.57.0.debian apt-cache policy sbcl sbcl: Installed: 2:1.0.57.0-1 Candidate: 2:1.0.57.0-1 Version table: *** 2:1.0.57.0-1 0 100 /var/lib/dpkg/status This is backported from the unstable sources, which for some reason didn't build on the Debian servers, so currently there isn't a version of this in unstable. CL-USER> *FEATURES* (CFFI-FEATURES:FLAT-NAMESPACE CFFI-FEATURES:X86 CFFI-FEATURES:UNIX :CFFI CFFI-SYS::FLAT-NAMESPACE :SWANK :SB-BSD-SOCKETS-ADDRINFO :ASDF2 :ASDF :ASDF-UNICODE :X86 :UNIX :ELF :LINUX :LARGEFILE :GENCGC :STACK-GROWS-DOWNWARD-NOT-UPWARD :C-STACK-IS-CONTROL-STACK :COMPARE-AND-SWAP-VOPS :UNWIND-TO-FRAME-AND-CALL-VOP :RAW-INSTANCE-INIT-VOPS :STACK-ALLOCATABLE-CLOSURES :STACK-ALLOCATABLE-VECTORS :STACK-ALLOCATABLE-LISTS :STACK-ALLOCATABLE-FIXED-OBJECTS :ALIEN-CALLBACKS :CYCLE-COUNTER :INLINE-CONSTANTS :MEMORY-BARRIER-VOPS :MULTIPLY-HIGH-VOPS :LINKAGE-TABLE :LITTLE-ENDIAN :OS-PROVIDES-DLOPEN :OS-PROVIDES-DLADDR :OS…

SBCL

Undecided / Fix Released

Notice the last entry.

Short summary - under certain circumstances, a garbage collection request does not do anything. Only reproducible under i386 though.

Fabby

8:13 AM

And if you worried about dumb questions, this is the kind of user we see over at AU:

0

Q: How to help a paranoid, delusional user?

Earlier today, this question was asked and contained a delusional rant about the KKK and the FBI and looked a bit like an episode of the X-files... My question is simple: Should we help this user (by calming him down as Andrew T. did) and concentrate on the technical side of the problem (as I d...

Faheem Mitha

@Fabby We get plenty of daft questions. Not so many from people who exhibit signs of mental illness.

Fabby

;-) That's why I asked the question...
BTW: an HVAC heats OR just ventilates OR does Airco and if you buy Mitsubishi, they're quite good in coastal environments as they are built more robustly.

Faheem Mitha

8:35 AM

@Fabby ok.

Hmm, an autotools question.

Faheem Mitha

9:58 AM

Hi @Caleb. New image?

Caleb

@FaheemMitha Yup. See explanation starting here.

Faheem Mitha

@Caleb ok

@Caleb Impressive artistic effort. How old is the child in question?

Caleb

@FaheemMitha 13 or 14.

Faheem Mitha

10:17 AM

@Caleb and that is supposed to be you riding the dragon?

Caleb

@FaheemMitha Yes.

Faheem Mitha

The dragon looks unhappy...

Is that a muzzle?

You work for a church? Or is that volunteer stuff?

Caleb

@FaheemMitha It's volunteer, but it's pretty close to full time volunteer work and I have my personal office from which I do other consulting/programming work to pay the bills setup at church so that I'm more readily available there.

Faheem Mitha

@Caleb Wow. Sounds busy. You do computer volunteer work, then? Being doing it long?

Caleb

@FaheemMitha Yes, and more. For > 10 years.

Faheem Mitha

10:26 AM

@Caleb I see. Interesting. You must enjoy it.

Faheem Mitha

1:38 PM

Can anyone tell me what the term "opencode" means here?

lwn.net/Articles/637745

derobert

3:46 PM

@FaheemMitha I'm not sure. Never seen that term before, and Google didn't help. All I could guess is he meant to say hardcode...

Faheem Mitha

@derobert Oh. I think I have seen that term somewhere before, but Googling didn't bring up anything.

Oh, you Googled too.

derobert

lkml.org/lkml/2005/4/22/40 is the closest I come up with via Google

Faheem Mitha

@derobert I don't immediately see the relevance ot this.

derobert

Well, simple way to find out—I left a comment asking.

At plus.google.com/+ArjanvandeVen/posts/VAK1SRHjTZm but I don't see a way to link directly to a comment.

@FaheemMitha "+Anthony DeRobertis I meant to indeed hardcode a specific algorithm by open coding a function using, say, low level functions from libcrypto/openssl rather than using the higher level primitives that those libraries provide that will pick crypto themselves." ... He answered.

terdon

4:14 PM

@Gilles are you around? Do you know how to specify repetitions in the regex flavor that's used by zmv? foo[a-z]{12} doesn't seem to work.

terdon

4:33 PM

AH, never mind, that's not about zmv, it's the shell itself that's doing the replacement.

terdon

5:09 PM

Perfect, thanks for answering @Gilles.

cuonglm

5:25 PM

@terdon: Are you there?

Braiam

is me cuonglm

cuonglm

@Braiam: What do you mean?

terdon

6:22 PM

@cuonglm I am now, but I will be leaving very soon. What do you need?

cuonglm

Did you re-read this unix.stackexchange.com/questions/192485/…

I updated my answer to make it more clearly :)

terdon

Oh, thanks! No, I hadn't seen the update (but I had upvoted the original). I have to go now, but I'll read it more closely when I get back.

Thank you :)

cuonglm

enjoy it :)

Faheem Mitha

7:31 PM

@derobert well, he again uses the term "open coding" without explicitly defining it. What do you think he means? Thanks for posting the comment.

Why do two people see different messages? And one is less clear!

Strange, why would you get a different message? Here is the complete text I see: "This question is protected to prevent "thanks!", "me too!", or spam answers by new users. To answer it, you must have earned at least 10 reputation on this site.". — Faheem Mitha 1 min ago

derobert

@FaheemMitha Not sure. Maybe you should only call low-level crypto functions after making sure the curtains are closed? Or after hiding under a desk?

But I think I've got the gist of his suggestion, even though I'm still not quite sure what open-coding means.

Faheem Mitha

@derobert kidding aside.

derobert

if you just ignore the word open it makes sense :-)

Faheem Mitha

@derobert Yes, we know what he thinks should be done, even though I at least am still not clear what "open coding" means.

@derobert I don't wanna.

:-)

derobert

@FaheemMitha Well, then you should take it to mean "coding in the open", and draw the curtains. Like I did.

Actually the blinds, but close enough.

More importantly, its time to go home.

Faheem Mitha

7:37 PM

@derobert Maybe he means you should code sitting in the park, rather than indoors. It's healthier. All that fresh air.

@derobert Early finish to the day?

derobert

@FaheemMitha But he recommends against opencoding.

And do they have fresh air in Mumbai, anyway?

Faheem Mitha

@derobert Oh, dammit. Fine, parks are dangerous places. There are dogs and things.

derobert

@FaheemMitha And kids throwing balls! They could hit your laptop.

Faheem Mitha

@derobert Not where I am. Depends what you call Mumbai, I suppose.

@derobert True. Those pesky kids.

No respect for coders.

@terdon any idea what happened to Alex (strugee)?

derobert

They'll keep bugging you if you're playing some video game you've never heard of too, as if there is nothing else you can do with a computer.

Faheem Mitha

7:40 PM

Haven't seen him here for awhile.

derobert

Anyway, I'm going home. Well, to Costco first, then home...

Faheem Mitha

@derobert Is this theoretical, or did it happen to you?

@derobert So short day? It looks like 3.30 pm EST, approximately.

I hear good things about Costco. I don't recall ever going, though.

derobert

@FaheemMitha Hah, it used to. I'm old enough now that people no longer presume my only use for computers is gaming.

Yep. It's Friday. No one else is here at the office. Time to leave.

Faheem Mitha

@derobert That's a relief.

@derobert Have a nice weekend.

derobert

I'll probably be around at least once.

Anyway, later...

Faheem Mitha

7:42 PM

Later

Faheem Mitha

7:54 PM

@Gilles why is this user seeing a different and less clear message?

When I access that post, it actually says exactly what I copied and pasted into my post above. — Nat Kuhn 22 mins ago

@terdon ^^

Faheem Mitha

8:31 PM

Nice post. Read it, everybody.

zgrimshell.github.io/my-journey-into-debian

Gilles

8:58 PM

> the strongest argument for some people advocating 4.0 seems to have been a wish to see 4.1.15 - because "that was the version of Linux skynet used for the T-800 terminator".

Faheem Mitha

"It's all just more of the same, just with smaller numbers so that I can do releases without having to take off my socks again."

/me wonders what socks have to do with anything.

""that was the version of Linux skynet used for the T-800 terminator""

Skynet used Linux? When/where?

Gilles

9:15 PM

@FaheemMitha reddit.com/r/linux/comments/2vyhr6/…

Faheem Mitha

I'm unclear what dh-systemd actually does. Does anyone know?

"It's systemd, evolved and gained awareness."

LOL

Gilles

@FaheemMitha Europeans count on their fingers by raising one finger at a time, which allows counting up to 10. (A thumb is a finger.) If you take off your socks, you can add toes to the series, for a total of 20.

Faheem Mitha

@Gilles Advanced Mathematics.

Thanks for the clarification.

@Gilles So, Linus is starting the 4.0 series. Clearly, the END IS NIGH.

Personally, I plan to sell out the human race if I am spared. Skynet, are you listening?

Does anyone know where that screenshot is from?

Faheem Mitha

10:29 PM

This proposal seems to be struggling. Shame, looks like a good idea.

77

Compilers

Proposed Q&A site for researchers and developers who want a serious and high-level answer on specialized topics on compiler architecture, e.g. parsing, type checking, intermediate code, optimizations, architecture specific optimizations and others.

Currently in commitment.

Faheem Mitha

10:53 PM

At the beginning of calibre-ebook.com/dynamic/calibre-usage it is written:

"Usage statistics are collected whenever a user starts calibre. Every calibre installation has a unique ID, this ID remains unchanged by upgrades and even an uninstall/re-install. This ID is used to collect usage statistics. Only this ID is stored, no other identifying information is collected."

Is there some way to watch this message being sent?

Markus

11:42 PM

evening

So if someone of you would like to collect some rep over at the raspberry exchange I would really appreciate if you could have a look at a (what I think is a) simple question: raspberrypi.stackexchange.com/questions/29087/…

Usually I hang around at stack overflow, in usual case you get answers within a couple of minutes, the activity on the rasberry exchange seem to be a little lower :)

Transcript for