The Whiteboard

1:26 AM

0

Q: Restricting object types that can be added to each other using the Composite Pattern

The Situation: I am building a custom PHP application framework. I have implemented a composite pattern so I can build a object tree representing the page to be rendered. Example: abstract class \Block\BlockAbstract { protected _blocks = array(); public function __construct($properties...

^--- Aside from the fact that my opinion is the basic premise is a bad idea, that's a very good question. Think I'll hazard coming up with an approach... If anybody has any ideas I'm curious to hear them

One immediate thought that comes to mind is a parser that generates the block types as a pre-compiler, since he's creating a type system for a runtime-typed language, and he wants good validation of the type system, creating a compiler-tool-thing that generates and or validates the block relationships before runtime might be a good approach

user55340

2:57 AM

I am not a fan of twitter, devops, or borat... that said... twitter.com/DEVOPS_BORAT

user55340

In startup we are practice Outage Driven Infrastructure.

Tweeted by DEVOPS_BORAT on March 12, 2013 at 4:38 AM

user55340

Devops good news! Devops is 100% peoples and culture so you not have of understand functional programming!

Tweeted by DEVOPS_BORAT on January 19, 2013 at 1:49 AM

user55340

In startup we are have new problem of deal with Big Metadata.

Tweeted by DEVOPS_BORAT on January 19, 2013 at 1:42 AM

user55340

When you are run DROP DATABASE you are have of many problem but Big Data is not of one of them.

Tweeted by DEVOPS_BORAT on October 25, 2012 at 3:50 AM

Ampt

4:32 AM

@MichaelT Still bitter. Hahhaa

user55340

4:42 AM

@Ampt Read some of that Devops borat and enjoy.

Jimmy Hoffa

1:54 PM

@MichaelT not a fan of devops?

user55340

3:22 PM

@JimmyHoffa most of the times I've seen things pro porting to be devops, its been "we don't have the QA or operational resources, so devs are going to do them too..."

user55340

Its one thing where everything is nice and small and, well, you've got a total of 5 people who are doing all of development, operations and QA in a startup environment.

Jimmy Hoffa

@MichaelT oh.. yeah that's a fair enough way of describing why we're moving that way... to be certain, we have IT and QA resources, we're trying to go towards devops because our resources are terrible...

but then we're not doing all of IT, just going to be managing our stuff

which still seems wrong to me..

user55340

The thing is, when you are at 'enterprise' sized products, the devops resources get streached way too thin.

Jimmy Hoffa

@MichaelT yeah, which we are...

we'll see what happens

that may happen, it's a risk we're aware of, we're planning on adding as much automation and instrumentation throughout the system as possible so hopefully it will be very easy for us to manage but who knows. We can always just say "Hey IT, you guys do it" if it fails, it's just that usually turns up more problems than us doing it around here..

user55340

The thing is, Dev and QA and Ops tend to have different perspectives and maintain a different what is working thought...

user55340

3:27 PM

Dev is "the code is right", QA is "the code is wrong - I'm going to find out how", and Ops is "The system is working, don't blame it... ok, I'll fix it"

user55340

Trying to put all three of those mindsets into a single team means something gets missed.

user55340

Dev testing their own code - tests the happy path. Something doesn't work? The system is broken, not the code.

user55340

In some environments, Devops is strictly forbidden. PCI for example - you can't have development environment and production environment mix. That means that your development machine can't connect to the production databases or servers. At all. That might be a bit extreme of a reading of PCI... but you get the idea.

user55340

Don't let the devs onto the production systems. Hot fixes are bad... those are things I want in places that touch money.

user55340

You really need to have four teams... Ops, QA, (new) Dev and (bug) Dev. Sometimes the two devs get mixed... but the bug dev is something that is the one that keeps the new dev from having to worry too much about context switches to old code lines.

user55340

3:35 PM

It sounds neat to have someone who does the hardware updates, keeps the bug tracker running, keeps the databases running, does the quick fixes, and writes the new code... but that means that none of those tasks gets the attention that they really need.

user55340

Its part of the 'cloud' world... hey, we put all our stuff in the cloud, now we don't need to worry about most of the database admin, or our servers... everything is now just writing code. But it isn't.. and so you've got devops. But its based on a mistaken assumption that everything in the cloud just 'works'

user55340

infoworld.com/t/application-development/…

user55340

>
Developers can end up being tasked with other roles, such as database administration, quality assurance, or testing, while the people in those roles are not going to be tasked with software development, says Knupp, a software engineer at online advertising company AppNexus. (He says in his own situation, he has a fine relationship with his company's operations team.) "What began as an experiment aimed at increasing software quality has become a farce, where the most talented employees are overworked (while doing less, less useful work) and lower-level positions simply don't exist." The ef…

Jimmy Hoffa

@MichaelT perhaps I misunderstand devops, I thought it just refered to devs having an active hand in the deployment/configuration/maintenance of the production environment and approaching it from a development standpoint of writing tools where valuable but focusing on automation etc so it minimizes the need for maintenance activities etc

We would never be involved in managing the IT stuff outside of just our server shit on production systems etc

user55340

Its one of those things like "QA" and "Agile". 95% of the places when they say "QA" they mean "tester"... and "Agile" means "just code it like a cowboy".

2

user55340

3:47 PM

The Dev part of software is often creative and innovative...

user55340

but QA and Operations need to be disciplined - QA isn't testing, its making sure good code is written and deployed through various processes and checks. Operations has auditing, and not giving everyone root on the production systems - if you really need it, use sudo... with the appropriate logging to make sure if you do mess it up, you can find out what you did and fix it.

user55340

Mixing the "free spirit" of dev into those areas - yes, the devs get a better view into it, but it also means that those often lose the discipline associated with their profession.

user55340

And so when I hear "devops" I hear "we're cutting back on QA and Ops, having the devs do them during the time they normally spend thinking about how to do the next thing that we don't really understand why they just stare at the screen."

user55340

Need a new database? Just go create it... you're devops. Bad data in production? You know it, fix it...

user55340

Its one thing to have devs be the third person you call when something goes wrong after operations has gone into it. Its quite another to have 'devops' be the only ones with the pager...

Jimmy Hoffa

4:37 PM

@MichaelT whoa yeah, I would not sign up for that shit.

user55340

@JimmyHoffa The ideal and reality of ${buzzword} are rarely the same.

user55340

Found an early project euler problem that fits very nicely into functional coding mindset.

user55340

> 2520 is the smallest number that can be divided by each of the numbers from 1 to 10 without any remainder.

What is the smallest positive number that is evenly divisible by all of the numbers from 1 to 20?

user55340

Its clear as recursion, map, and reduce in my mind...

user55340

Take the range 10..12 as an example. Find the prime factors of each of the numbers and map them: 10 .. 12 => {2=>1, 5=>1}, {11=>1}, {2=>2, 3=>1}. Then reduce that list to the max for each power: {2=>2, 3=>1, 5=>1, 11=>1}

ratchet freak

4:59 PM

that is just multiplying 2520 with 11,13,2,17 and 19 (all of the primes and the extra 2 so it divides 16 )

Jimmy Hoffa

knowyourmeme.com/memes/48293

nonsense

    Says "48÷2(9+3) = " is an unknown unsolvable equation! Bullshit.
    It's the same equation as this, and this is a very obvious and solvable equation:

       48
    ______

     2(9+3)

ratchet freak

it's just badly written

and unclear whether it means (42/2)*(9+3) or 42/(2*(9+3))

that's the only issue with that

Jimmy Hoffa

@ratchetfreak doesn't matter, the thing that frustrates me is there's a bunch of people saying "O see math doesn't have a solution for this very simple equation!" nonsense, math easily has the answer

the equation can be rewritten in the form I said and it's an identical equation with no change in precedence rules or additional parenthesized terms etc, it just makes clearer what the precedence rules are

but those precedence rules are defined and are known and people who say they're not are just trying to show how confusing math is which helps no one by making faux confusion

ratchet freak

just like the troll proofs of 0=1

gnat

5:32 PM

@MichaelT after giving a bit more thought to your "theory" about that decorator guy, I think it explains my observations fairly well

user55340

5:50 PM

@gnat The problem with the questions he asks is that he really needs to sit down and do the fundamentals, but that can't be an answer to the questions he's asking.

user55340

@ratchetfreak Yep, but the thing is to write some nice FP code that does that correctly (partly so I can get my head around map / reduce).

Jimmy Hoffa

all the numbers from 1 to 20
let n = [1..20]
smallest positive number
let p = [1..]
that is evenly divisible
let remainderless x y = mod x y == 0
by all of the numbers n
head $ dropWhile id $ map (\x -> and $ map (remainderless x) p) n

no reduction necessary, iterate from 1 forward, stop dropping things when you get a true, take the head from there and you have the smallest because you iterated up from the bottom

gnat

@MichaelT yes indeed. Hits "Ask Question" too early...

151

A: How much research effort is expected of Stack Overflow users?

A lot. An absurd amount. More than you think you are capable of. In fact, asking a question on Stack Overflow is the absolute last thing you ever want to do. You want to avoid it at all costs. You want to think of it as a horrible shame1 that will forever haunt you and pass down from you to your ...

Jimmy Hoffa

6:08 PM

actually that just get's you the first true, but I forgot to carry the value along with it. Correctly it should be:

snd.head $ dropWhile (not.fst) $ map (\x -> and $ map (\y -> (remainderless x y, y)) p) n

@MichaelT $ means parenthesize everything on the right side of this point

obviously that's naive though

it's a full iteration solution with no culling