terrible programmers: my god, it's full of chars

The Something Awful Forums > Discussion > Serious Hardware/Software Crap > YOSPOS > terrible programmers: my god, it's full of chars

«‹›1446 »

the talent deficit: Dec 20, 2003; self-deprecation is a very british trait, and problems can arise when the british attempt to do so with a foreign culture

could someone explain to a bad programmer why i shouldn't be concerned that like basically every distributed platform that isn't yarn (possibly even yarn) relies on zookeeper? isn't that a single point of failure and a potential bottleneck? is zookeeper that reliable?

# ? Sep 28, 2014 08:19

Adbot: ADBOT LOVES YOU

# ? May 11, 2024 12:04

Quebec Bagnet: Apr 28, 2009; mess with the honk
you get the bonk; Lipstick Apathy

~Coxy posted:

don't mock out your database. some people say that this means it's not proper Unit Testing anymore but who gives a gently caress

well, it's not if you have actual logic relating to your database. for example, if you have stored procedures of any kind, that's another piece of code that could have a problem, so you don't know if you're testing the object you want to test or how it interacts with your specific database implementation

# ? Sep 28, 2014 08:20

Quebec Bagnet: Apr 28, 2009; mess with the honk
you get the bonk; Lipstick Apathy

btw Bloody you should look into bcp.exe for loading you're data quickly

# ? Sep 28, 2014 08:22

DONT THREAD ON ME: Oct 1, 2002; by Nyc_Tattoo; Floss Finder

BONGHITZ posted:

c#f#r#

what is next im going crazy

# ? Sep 28, 2014 12:04

coffeetable: Feb 5, 2006; TELL ME AGAIN HOW GREAT BRITAIN WOULD BE IF IT WAS RULED BY THE MERCILESS JACKBOOT OF PRINCE CHARLES

YES I DO TALK TO PLANTS ACTUALLY

BONGHITZ posted:

c#f#r#

what is next im going crazy

http://research.microsoft.com/en-us/projects/fstar/

# ? Sep 28, 2014 12:32

Soricidus: Oct 21, 2010; freedom-hating statist shill

F*** you

# ? Sep 28, 2014 12:39

suffix: Jul 27, 2013; Wheeee!

the talent deficit posted:

could someone explain to a bad programmer why i shouldn't be concerned that like basically every distributed platform that isn't yarn (possibly even yarn) relies on zookeeper? isn't that a single point of failure and a potential bottleneck? is zookeeper that reliable?

1. zookeeper is very reliable

2. you can and should have several zookeeper servers, e.g. if you have 5 servers in a cluster you can lose any 2 of them

3. you should be more concerned about HA setups that don't use zookeeper, since that means they're writing their own consensus protocol, which is very tricky to get right

that said, yes a botched upgrade can take down everything, and zookeeper can become a performance bottleneck if you do a lot of writing
if you're really big and care about this, each service can have its own zookeeper cluster

zookeeper has some faults, but it's a known quantity and can prevent you from being a mongodb or an elasticsearch

# ? Sep 28, 2014 13:12

MononcQc: May 29, 2007

the talent deficit posted:

could someone explain to a bad programmer why i shouldn't be concerned that like basically every distributed platform that isn't yarn (possibly even yarn) relies on zookeeper? isn't that a single point of failure and a potential bottleneck? is zookeeper that reliable?

So Zookeeper is that very solid place in your system where data can be safely serialized. Zookeeper has good consistency guarantees through its distribution protocol (ZAB^PDF), which ensures safe replication and implicit leader election.

This means Zookeeper can have 3 or 5 nodes (more than that and it starts being a lot slower to replicate, and therefore to write anything) and see a minority of them fail and keep being available. This is a quorum system, equivalent to Paxos or Raft in capacity; In your system, zookeeper is used as the main point of serialization for writes and only requires a majority to work, but has better availability properties than a SQL database by virtue of self-electing a new leader without data loss.

Also, given it works like other Quorum systems, you can have 'learner' or 'observer' nodes, which replicate state, but do not ever write to it -- they can be queried and used in a way to scale up by only using the voting nodes as writers, and the learners for reading, unloading the writers a bit.

So for systems that tend to:

Have writes that can fit a single node
Tend to have more reads than writes
Can afford to pay for some redundancy to cancel out individual node failure
Can deal with something closer to K/V stores than full blown SQL transactions

Zookeeper is a very good choice, and a shitload more mature than stuff like etcd (using Raft).

E: because Paxos is a quorum system, it still lies on the CP side of CAP -- it can just cope with a minority of losses, which is nicer than SQLs, also on CP side of CAP but unable to easily deal with the loss of the only master they have. Quorum is generally the most available side of CP you can get, but can still freeze everything if you lose a majority of your writers, or if they get stuck in netsplits that leaves no clear majority

MononcQc fucked around with this message at 13:24 on Sep 28, 2014

# ? Sep 28, 2014 13:20

serious norman: Dec 13, 2007; im pickle rick!!!!

im a Delphi guru

# ? Sep 28, 2014 14:32

~Coxy: Dec 9, 2003; R.I.P. Inter-OS Sass - b.2000AD d.2003AD

chmods please posted:

well, it's not if you have actual logic relating to your database. for example, if you have stored procedures of any kind, that's another piece of code that could have a problem, so you don't know if you're testing the object you want to test or how it interacts with your specific database implementation

who cares

i test it and it works, or it doesn't.

# ? Sep 28, 2014 14:35

tef: May 30, 2004; -> some l-system crap ->

MononcQc posted:

The advantages of TDD are:

- You think about what your code should do and expose as an interface before jumping in (write it with the user in mind, not the programmer)
- Your tests are more about what the code should do than crystallizing what crappy code you wrote is doing right now
- It's a shitload more boring to write code after everything is done, than making it part of the design process

- You don't think about your code that much beyond can I double everything going in.
- Sometimes the tests you write are so divorced from the code you can write tests that pass even if the code doesn't exist
- A series of crappy tests give you a false sense of security, but in reality give you a maintenance burden to changing the api

The thing about TDD is you end up with some tests, which sometimes, is better than nothing. Some people find that test-first makes them think about design, or helps them focus. Other times you end up with a bunch of sleep(0.1) statements because life is suffering.

TDD without thinking is some of the test driven i've encountered. It's worth trying test first, along with various other strategies. It isn't worth doing test first if you're going to write lovely tests � TDD means you will have more tests but it does not guarantee better design for code. It isn't a magic wand and you should kill any buddha who claims otherwise.

If you're asking if you should do it, the answer is probably yes.

It is probably worth doing because you'll expose yourself to more mistakes quickly, and making lots of mistakes afaik is the best way to learn programming.

# ? Sep 28, 2014 14:51

tef: May 30, 2004; -> some l-system crap ->

I'm not against testing, or strong typing, but they don't come for free, and don't guarantee good, performant, robust software.

However it is unlikely your code is good if you don't check your assumptions, and you're probably not checking them if you don't have code to check them for you. Testing does not give quality, but the quality of your tests will impact the quality of your software.

I kinda want more people to use things like dialyser and quickcheck :3:

# ? Sep 28, 2014 14:55

Jabor: Jul 16, 2010; #1 Loser at SpaceChem

you should almost certainly try tdd if you haven't yet

you don't have to stick with it after you've tried it tho

# ? Sep 28, 2014 15:10

kitten emergency: Jan 13, 2008; get meow this wack-ass crystal prison

is there a good book or something on tdd

# ? Sep 28, 2014 15:17

MononcQc: May 29, 2007

uncurable mlady posted:

is there a good book or something on tdd

Never even give Beck's book a read. I hated it and it's the cargo-cult TDD bullshit where you write tests instead of thinking about your problem. I much preferred Growing Object-Oriented Software, Guided by Tests, though it's not like it's a super entertaining read.

# ? Sep 28, 2014 15:22

kitten emergency: Jan 13, 2008; get meow this wack-ass crystal prison

thanks

# ? Sep 28, 2014 15:32

Malcolm XML: Aug 8, 2009; I always knew it would end like ｔｈｉｓ．

tests are basically an executable spec, they limit the space in which your program can grow, but dont limit it to much or u end up straightjacketing urself

hth

# ? Sep 28, 2014 15:45

qntm: Jun 17, 2009

Jabor posted:

you should almost certainly try tdd if you haven't yet

you don't have to stick with it after you've tried it tho

yeah absolutely, it is far from a silver bullet but everybody should know how to do it and when

fixing defects being the big one

# ? Sep 28, 2014 16:36

Valeyard: Mar 30, 2012; Grimey Drawer

chmods please posted:

btw Bloody you should look into bcp.exe for loading you're data quickly

# ? Sep 28, 2014 18:00

master of the sea: Apr 16, 2003; *skweeeeeee*

just found engineering mathematics 7th edition on ebay, soo great, i am finally going to conquer my god drat crippling math disability.

# ? Sep 28, 2014 21:19

Squinty Applebottom: Jan 1, 2013

mathematics is a socially transmitted disease

# ? Sep 28, 2014 21:20

fritz: Jul 26, 2003

Squinty Applebottom posted:

mathematics is a socially transmitted disease

not that socially

# ? Sep 28, 2014 21:29

theadder: Dec 30, 2011

sexually

# ? Sep 28, 2014 21:31

Notorious b.s.d.: Jan 25, 2003; by Reene

theadder posted:

sexually

that thought makes erdos numbers chilling

# ? Sep 28, 2014 21:50

jony neuemonic: Nov 13, 2009

uncurable mlady posted:

is there a good book or something on tdd

this one is pretty good if you work with python or on web stuff.

# ? Sep 28, 2014 22:02

Symbolic Butt: Mar 22, 2009; (_!_); Buglord

Obey the Testing Goat

# ? Sep 28, 2014 22:04

jony neuemonic: Nov 13, 2009

Symbolic Butt posted:

Obey the Testing Goat

web developers won't write tests unless you make it so wacky!!

# ? Sep 28, 2014 22:08

coffeetable: Feb 5, 2006; TELL ME AGAIN HOW GREAT BRITAIN WOULD BE IF IT WAS RULED BY THE MERCILESS JACKBOOT OF PRINCE CHARLES

YES I DO TALK TO PLANTS ACTUALLY

master of the sea posted:

just found engineering mathematics 7th edition on ebay, soo great, i am finally going to conquer my god drat crippling math disability.

it wont be easy or quick but its mos def worth it

# ? Sep 28, 2014 22:51

master of the sea: Apr 16, 2003; *skweeeeeee*

coffeetable posted:

it wont be easy or quick but its mos def worth it

yes that's what I figure i might as well just grind this out once and for all

# ? Sep 28, 2014 23:33

BONGHITZ: Jan 1, 1970

when you die another human will have to be trained in math to replace you

# ? Sep 29, 2014 00:22

Kathleen: Feb 26, 2013; Grimey Drawer

coffeetable posted:

http://research.microsoft.com/en-us/projects/fstar/

oooh this is cool. it's good to see microsoft knows how important .net is to their long-term survival

# ? Sep 29, 2014 14:05

Pittsburgh Fentanyl Cloud: Apr 7, 2003

Luigi Thirty posted:

i have only ever done this and web things which is why uis are stupid and confusing

don't sign your posts lmfao

# ? Sep 29, 2014 14:33

gonadic io: Feb 16, 2011; >>=

I'm playing around with http://rise4fun.com/FStar/tutorial/guide and I'm trying to write a tail recursive reverse function on vectors.

I have:

code:

module ILIST0

type ilist : int => S => S = 
   | INil  : ilist 0 'a
   | ICons : x:'a -> n:int -> ilist n 'a -> ilist (n + 1) 'a

val reverse : n:int -> ilist n 'a -> ilist n 'a
let reverse n l =
  let rec reverse2
            (n1:int)
            (acc:ilist n1 'a)
            (l':ilist (n-n1) 'a)
             = match l' with
    | INil -> acc
    | ICons x ln_minus_1 ls -> 
      reverse2 (n1+1) (ICons x n1 acc) ln_minus_1 ls
 in reverse2 0 INil n l

and am having a number of problems: since I can't figure out how to write an explicit type sig for reverse2 and slip a {n1+n2=n} (where n2 is the length of the second list), I'm instead having to do it by pattern matching on the length of the second list as (n-n1) and hope it works out what I mean. The error I'm getting is:

input(13,15-17,3) : Error : Expected a pattern of type
ILIST0.ilist (Sub (x_8) (x_10_2)) 'x_7_2
But got pattern of type
ILIST0.ilist 0 '_U309
Type checking failed: ILIST0

which seems to me that it can't prove that the length of the list resulting from reverse2 will be n

# ? Sep 29, 2014 15:12

gonadic io: Feb 16, 2011; >>=

nevermind, i got it working by specifying reverse2 on its own so i could give it a proper type sig. is there a way to do a typesig inside a let binding inside a function?

# ? Sep 29, 2014 15:21

The Leck: Feb 27, 2001

gently caress them posted:

Leaving my comfortable little womb of kushiness with nonexistent standards or deadlines and no technical coworkers is the obvious first step. There is nothing to keep me here accountable except shame or a desire to improve. I could be the laziest fucker ever, and retire comfortably. gently caress that.

are you me? this is worryingly accurate.

# ? Sep 29, 2014 19:04

Janitor Prime: Jan 22, 2004; PC LOAD LETTER

What da fuck does that mean; Fun Shoe

The Leck posted:

are you me? this is worryingly accurate.

god drat it same

# ? Sep 29, 2014 19:33

Bloody: Mar 3, 2013

this is unrelated to my previous line of questions

say i have a c# program, butt, that produces objects, farts, and i have another (or several other) c# programs on my network that consume farts and return odors and i want the butt to produce farts and get back odors via these other programs over the network (because the odor of a fart takes a lot of cpu time to figure out)

how do i do this

# ? Sep 29, 2014 19:40