The Whiteboard

Lightness Races in Orbit

12:16 AM

Ixrec

1:55 AM

"Our system has identified this post as possible spam; please review carefully"

that's new

anyone know what makes that message appear in the Close Votes queue?

ratchet freak

spam flag or a questionable link

I think

Duga

2:28 AM

Overall development process is off-topic/too broad for SO. Check if rules of programmers.stackexchange.com permit this question. — Alexei Levenkov 54 secs ago

user55340

Hmm... only two perl6 questions on Code Review. This must be fixed! (hint to: @amon)

Mark Cramer

3:26 AM

hello?

this is my first time here. am i alone? is this space for just chatting about programming?

ratchet freak

4:12 AM

there's usually more people here during the american working day

ambigram_maker

8:06 AM

Hello guys... I need your opinion on a mathematics library that I want to make in Java. Here is the link to the Google Form.

Ixrec

11:18 AM

@ambigram_maker I don't think that's an answerable question a priori, what are your main use cases?

most of the questions in that survey boil down to "What would your target audience and their use cases want it to do?" and that's probably not us (at least, I have no burning need for a new arbitrary-precision arithmetic library)

Duga

12:21 PM

Maybe post this on CodeReview.SE or Programmers.SE? I'm not particularly sure which site is most suited to this kind of question. — Arc676 12 secs ago

Lightness Races in Orbit

12:32 PM

you know who I hate? Chrome extension developers. 20MB RAM per piece! WHYYY?!!?

"you need to buy more RAM" "no you twit you need to stop using so much of what I have already"

> Some people might say it is "Reinventing the Wheel", but I assure you... my approach will be slightly different from the existing ones.

lol

Jimmy Hoffa

1:53 PM

@LightnessRacesinOrbit said everyone ever who reinvented the wheel

> Draper is renowned for his lifelong [...] and his poor personal hygiene.

Man, when that's right in your wikipedia page, you need to take a damn bath

user55340

3:18 PM

Ha! FGITW eat my shorts.

user55340

2

Q: Why is the Git .git/objects/ folder subdivided in many SHA-prefix folders?

Git internally stores objects (Blobs, trees) in the .git/objects/ folder. Each object can be referenced by a SHA1 hash that is computed from the contents of the object. However, Objects are not stored inside the .git/objects/ folder directly. Instead, each object is stored inside a folder that s...

user55340

2

A: Why is the Git .git/objects/ folder subdivided in many SHA-prefix folders?

It is possible to put all the files in one directory, though sometimes that can become a bit large. Many file systems have a limit. You want to put a git repository on a FAT32 formatted drive on a USB stick? You can only store 65,535 files in a single directory. This means that it is necessary...

user55340

(not a short answer)

user55340

0

A: Why is the Git .git/objects/ folder subdivided in many SHA-prefix folders?

There are some filesystems and/or filesystem implementations and/or libc implementations where performance degrades with large numbers of directory entries.

user55340

0

A: Why is the Git .git/objects/ folder subdivided in many SHA-prefix folders?

These 256 bucket allow git to store larger repositories on file systems that limit the number files in a directory and provide descent performance on file systems that become slower with directories containing many files.

user55340

3:19 PM

Mine was 15:16:07, Jörg was 15:16:39, Kasper was 15:16:35

Ixrec

have another upvote

Jimmy Hoffa

True horror: When your 5 year old has his own bathroom.

Happy Halloween.

Lightness Races in Orbit

...

your life will be horrible when you don't have enough bathrooms to go around for each o your children

i.e. you have more children

such a nice day and I'm not going anywhere sigh

overexchange

4:07 PM

Hai

Some good news.. I am nominated for Saskatchewan province this week. I would further apply to CIC to approve my PR.

I can move out to other province, if I do not get a job in SK province.

Lightness Races in Orbit

Great!

Belated happy birthday, btw :)

user55340

5:13 PM

Cats ARE neurotic - and they're probably also trying to work out how to kill you, say researchers

user55340

> Only their size prevents the cat from being able to unleash its full predatory qualities.

Ixrec

that article is amazingly lacking in detail

user55340

However, one must realize that this was posted on halloween... and such things as this are posted to image sharing sites...

user55340

Yes, your cat is trying to figure out how to kill you.

user55340

5:16 PM

user55340

See? Ferocious predator about to kill something.

Ixrec

@MichaelT what character is that supposed to be?

user55340

@Ixrec Daryl

user114359

5:40 PM

@MichaelT How to tell if your cat is plotting to kill you

user55340

Never Turn Your Back on BIG CATS!

►

user114359

@MichaelT well yeah, big predators are going to do that. But a house cat is more likely to set a trap using a belly rub as a lure: it ends in teeth, claws, and agony.

user55340

@Snowman breakingcatnews.com/comic/…

user114359

Sometimes, I really just want to comment:

user114359

-5

Q: Syntax error print

I'm getting a syntax error while trying to follow an infinite skills tutorial. If this is because I'm using a newer version than the instructor is, how can I keep using mine and prevent this in the future? This is the code: print ("hello,", end=' ') print (" world!") I get a syntax error for ...

syntax errors python-3.x

Ixrec

5:55 PM

I'm not even sure what to comment on that question

any form of "you need to read the tutorial a lot more carefully" is just going to sound rude, and answering the question directly would basically involve explaining lexing/parsing

user114359

It is off-topic anyway, which is why I typed that simple custom close reason.

user114359

cross-posted:

user114359

-2

Q: Syntax error in single line

I'm getting a syntax error while trying to follow an infinite skills tutorial. If this is because I'm using a newer version than the instructor is, how can I keep using mine and prevent this in the future? This is the code print ("hello,", end=' ') print (" world!") I get a syntax error for the...

python

user55340

6:55 PM

Yep... quality content right there (in the history - its not exactly safe for work).

user55340

And on locked questions... can't work to curate it at all (not that its easy to curate a question with 180 answers)

Lightness Races in Orbit

7:09 PM

That is NOT a surety! People don't read guides on how to ask questions naturally! But for the sake of the stack exchange 'gods' I'm gonna be a good idiot and try to learn how to pose a question based on being told what to think. HAVE A NICE DAY. — Jake Cook 1 hour ago

lol

umad

user114359

@LightnessRacesinOrbit ragequit

Lightness Races in Orbit

huh, if I build a bunch of VC++ and VB.NET projects on Linux under Mono, can I just run those unmodified on arbitrary Windows PCs (assuming prior installation of the typical .NET runtimes)?

then I don't need a Windows build server or VM

amon

7:28 PM

@LightnessRacesinOrbit If you deploy on a platform, then test on that platform.

More accurately, if price_of_license/your_hourly_rate + time_spent_setting_up_vm - time_spent_debugging_platform_specific_issues could reasonably drop below zero during the next couple of years, then setting up a CI environment on Windows makes economic sense. It probably does.

Lightness Races in Orbit

true in general

unlikely in our case

with some exceptions (mostly software that's now dead, though may make a resurgence in future), I'm really only talking about internal tools that few other people seem to care about (even though I'm quite proud of them)

I couldn't get company time to create them, let alone set up CI for them :P

just thought if I can technically achieve it, I can whip up a Docker container and be done with it, obtaining at least basic stability checking through my shiny new Jenkins installation (not to mention convenient on-demand builds)

eh well you're probably right; a dangerous path to travel regardless

(for sellables, our pre-release testing would of course still take place on the target platform. the fact that these are GUIs, and the build server is headless, and I'm trying to keep windowing servers as far away from it as possible, certainly helps with that)

MetaFight

7:43 PM

7:45pm instant coffee :|

user55340

@MetaFight thecodelesscode.com/case/106

Lightness Races in Orbit

@MetaFight ew

MetaFight

@MichaelT so what's the moral of the story there? Never talk about failures or victories?

user55340

@MetaFight git blame is where all the information is stored.

MetaFight

@MichaelT Keep in mind that I'm only on my first cup of instant. I don't smell what you're cooking.

also, I've never used git blame.

user55340

7:54 PM

Ok, svn blame.

MetaFight

I've never used <source control> blame. What is it?

user55340

svn blame (praise, annotate, ann) — Show author and revision information inline for the specified files or URLs.

Ixrec

if you happen to know that a particular line of code is causing problems of some kind, * blame tells you when that line was most recently changed and by whom

MetaFight

@MichaelT ok, yeah, so the source control system records who actually did the relevant work. I'm still not sure what the takeaway of that blog post was. Sometimes it's a good thing to share your victories... especially when they're rare in the team.

user114359

Basically it shows the most recent commit for each line. If I mouse over a line, all other lines in that commit (that don't have another more recent change) light up and the commit info shows in a popup. At least this is how the Eclipse SVN plugin does it.

user55340

7:56 PM

While sharing victories is good, becoming the team hero is bad.

MetaFight

Well, at least in my previous job, it was either be the hero, or watch the project go nowhere.

user55340

c2.com/cgi/wiki?HeroicProgramming

MetaFight

Explaining the problems to my team members and then trying to develop solutions with them never worked because they didn't care.

@MichaelT I completely agree. That's why I left that place.

user55340

November marks the birthday of Charles Schulz, pioneer of tongue awareness.

MetaFight

I sometimes wonder if I care about progress, process, and efficiency too much. I'm currently part of a volunteer organisation that's going through some pretty big changes. I've spotted some pretty inefficient processes and have spoken to a few people about ways to turn things around. Most people seem to think I'm just uptight. I personally don't like seeing something that should be a 1 day task take a full 3 weeks... but that seems to be acceptable for everyone else.

I'm happy to accept this more relaxed approach because it's just volunteer work. In the office, on the other hand, I would be driven insane! :)

user114359

8:05 PM

@MetaFight don't forget that those types of organizations are not profit-driven, which removes a large motivation to increase efficiency. Businesses know that more efficiency == more profit (in general).

user114359

They also tend to have leadership that is in place for long tenures, and "efficiency" could be seen as "some young punk doesn't like MY way of doing things" and (see my previous point) there is no argument to be made that change will benefit them.

user114359

i.e. nonprofit leadership can be very territorial.

MetaFight

@Snowman yeah, I know. I have to keep reminding myself it's not a job and that people only volunteer if the work is still rewarding and fun. I'm currently between jobs and am constantly treating the volunteer work like an actual job (because I've got so much free time). I'll get used to it soon.

In this organisation most of the work is done by a single person. That worked fine originally, but nowadays there's way too much work for just one person. I need to find a way to help out without threatening the existing powers that be.

communication/social interactions == hard stuff

Ixrec

8:21 PM

deleting master on purpose feels really weird

user114359

@Ixrec ever do rm -rf / on purpose?

Ixrec

definitely not on /

on /build/, absolutely

user114359

I did that once before I formatted and reinstalled just to see if it would work. Modern rm versions have protections built-in, but back in 1998 or so it almost worked

Lightness Races in Orbit

a bit like me

the other day

I almost worked

user114359

it couldn't remove everything, but it did manage to delete /bin so I couldn't even ls anymore

Lightness Races in Orbit

8:24 PM

I sort of did it by mistake once but it wasn't really my fault

I had a brand new dedi ready to configure, and I was almost there, about to set into motion my careful failover plans, domain TTLs all nice and low, mail server being mirrored, all proud of myself.

There was this one odd folder that I couldn't explain, that seemed to be an accidental copy of parts of the root filesystem. I figured, huh, that shouldn't be there, and deleted it.

Of course it turned out that somehow the filesystem mounts had gone wrong, and it actually was parts of the root filesystem, mirrored from / into /something/somethingelse.

Bye bin.

I cancelled my plans, put everything back to normal on my original server, let the contract lapse on the new one after the first month and never bothered migrating again.

Saved myself a couple of quid each month I guess

(the dedi was marginally more expensive than my VPS)

user55340

8:44 PM

For fun, try this: find -perm +X -exec {} strip; - saves a bit of space by removing debugging symbols from executables.

user55340

Note: the kernel is executable.

user55340

Further note: the debugging symbols are also used for dynamic linking.

Lightness Races in Orbit

lol

are they?!

user55340

Back in college, one of the sysadmins did that once. Remember, we're talking things like "there's only 20 megabytes for the system and users on this volume.... lets see what space we can clean up?"

user55340

And when you start stripping the executables from your home directory... you can save a few kilobytes. Thats a lot on those 20 megabyte hard drives.

user55340

8:47 PM

Ooops... there's the shared libraries and oops... the kernel is borked now too.

user55340

The only things that worked were the statically compiled binaries.

user55340

Had to restore from backup on that system.

user55340

Frustrating part - a lot of things almost worked.

Lightness Races in Orbit

why are debug symbols used for dynamic linking o.O

user55340

Another fun one (this was from SGI tech support) was a guy who didn't like having to log in as a user... so he did a chmod -R 777

Lightness Races in Orbit

8:49 PM

heh

user55340

that removed the set uid bits and he could only log in as root afterwards (because login couldn't do the 'change to another user' unless it was run by root)

Ixrec

why did the system let him chmod anything when he wasn't logged in as anyone?

Lightness Races in Orbit

he was logged in

he didn't like it

Ixrec

derp, correction

user55340

Debug symbols are things like "the function XYZ is at location 0x01234568 in the shared library" which is what debuggers and dynamic linkers like to know.

Ixrec

8:50 PM

why did the system let him do anything at all without being logged in

Lightness Races in Orbit

oh, those kinds of debug symbols

no, wait

bluh I'm confused

hrm

surely you can strip debug symbols. you just can't strip them for any exported symbols.

at that point I'm not sure those are really called debug symbols any more

user55340

9:08 PM

It looks like the modern 'strip' command is a bit more intelligent...

user55340

--strip-all
Remove all symbols.
-g
-S
-d
--strip-debug
Remove debugging symbols only.

user55340

This was back in '92 on some version of HP-UX.

Lightness Races in Orbit

aha

Ixrec

9:46 PM

so I'm trying to understand the whole "vendoring" debate, i.e. I want reproducible builds, therefore I must have copies of my dependencies in my source tree, but how to manage those copies without creating additional headaches

specifically I've read about submodules and subtrees in git, and...I'm completely failing to see what advantages they have over straight-up "manual" vendoring, where I just copy-paste into my repo

anyone here more familiar with how those features are supposed to work?

Lightness Races in Orbit

> therefore I must have copies of my dependencies in my source tree

weeerrrrrrr what

please god no

Ixrec

what's the alternative?

I saw a few rants online to that effect but they all either never addressed how you're supposed to get reproducible builds or they said "use submodules instead" and I don't understand why that's supposed to be better

Lightness Races in Orbit

um document dependencies

if you bundle everything into your source tree then they're not dependencies any more

they are part of your project

redundant ones at that

enderland

11:09 PM

@Ixrec how does your build work currently?

enderland

11:22 PM

also @LightnessRacesinOrbit do you ever sleep? aren't you in Europe?

Ixrec

11:50 PM

@enderland at the moment this is just a hypothetical question since I don't have any projects at work where vendoring would improve build repeatability (it's usually other code within our company that breaks us...)

user55340

@LightnessRacesinOrbit look at the revisions on that question: stackoverflow.com/posts/5508110/revisions

Ixrec

but I'm thinking primarily of future hobby projects I might do in C++ (i.e., one of the languages without a proper, widely-used packaging system that makes "documenting" the dependencies a complete solution)

user55340

my favorite line is:

user55340

> Post Deleted by Diego Mijelshon, Sklivvz♦, yoda, bmargulies, John Saunders, Kyralessa, ChrisF♦, meagar♦, George Stocker♦, Anna Lear♦

enderland

lol

user55340

11:53 PM

Will Ferrell and Chad Smith Drum-Off

►

Ixrec

in the past I had a hobby project which used SFML (a cross-platform C++ lib for basic sound, graphics, etc) and I could never seem to work out a good way of keeping its source repo alongside my source repo without things getting very fiddly

Transcript for