The Whiteboard

Ixrec

2:00 PM

oddly enough I remember one of my dad's electric shavers having a Java logo on it...

bigcodeszzer

my prof has a software suite that lets you compile c++ code and then run it on an embedded system.

I'd probably use that if I was programming something like that

BarryTheHatchet

usually one would use C for embedded because the overhead of the typical C++ runtimes is so high

Ixrec

depending on what embedded means that may demand just plain C

bigcodeszzer

compiled into machine

BarryTheHatchet

but for what I call "pseudo-embedded" that isn't really embedded, like Raspberry Pi, you can just use a bog standard ARM C++ cross-compiler with all the bells and whistles

Ixrec

2:01 PM

CBS why u no let me watch Supergirl

bigcodeszzer

then put directly onto

Ixrec

fix your video player app thing

BarryTheHatchet

licensing innit

they no let us cos no usa

Ixrec

nah I'm VPNing as usual

and it lets me onto the page with the episode

bigcodeszzer

Anyways. I don't think there is overhead

Ixrec

2:01 PM

it just doesn't load

outside the VPN it correctly shows me an error message

bigcodeszzer

There is a fair wait time though I think.

BarryTheHatchet

@bigcodeszzer well, you're wrong

C++ runtime libraries are quite large

bigcodeszzer

It's compiling it into machine code

Ixrec

the C++ features that aren't in C definitely have overhead that's noticeable on very resource-constrained systems

bigcodeszzer

And the machine code goes on the imbedded system

BarryTheHatchet

2:02 PM

so? a little microcontroller still doesn't have the space for that

Ixrec

especially since C++ requires a runtime

does C require one? I can never remember

BarryTheHatchet

or the power to run it, most likely

bigcodeszzer

Well

BarryTheHatchet

@Ixrec you can get by without libc but it's often small enough anyway

bigcodeszzer

The guy was running it on a miniscule chip

Ixrec

2:03 PM

yeah I'd assume libc is much much smaller than libc++

not physical size lol

bigcodeszzer

I know

my point is

BarryTheHatchet

I thought you wanted to talk about exceptions

bigcodeszzer

He had the thing working on a small chip

I did.

Ixrec

@BarryTheHatchet I'm starting to wonder that myself

bigcodeszzer

Anyways. Yes. Exceptions. In C++

Why do you use them

BarryTheHatchet

2:04 PM

@Ixrec I think perhaps my chatroom spidey sense is more finely honed than yours.

Ixrec

because some failures are unavoidable and error codes are often much worse

BarryTheHatchet

(no offence)

Ixrec

@BarryTheHatchet well I was wondering it a looooong time ago

bigcodeszzer

But can you demonstrate that with an example?

that's why I was directed here

Ixrec

memory allocation failure is by far the most compelling example

if that one doesn't convince you I'm not sure what else to turn to

bigcodeszzer

2:05 PM

But why not just use an if()

One we used was opening a file

Ixrec

around every single memory allocation call in your entire program?

BarryTheHatchet

Yes, come to chat as I have a good example that I am too lazy to make into an actual answer. — whatsisname 14 hours ago

but @whatsisname is not here.

Ixrec

you know most C++ memory allocations are not explicitly performed by your code like a malloc() call would be, right?

bigcodeszzer

Where do you use the try catch?

That is catching those?

BarryTheHatchet

4265

Q: The Definitive C++ Book Guide and List

This question attempts to collect the few pearls among the dozens of bad C++ books that are published every year. Unlike many other programming languages, which are often picked up on the go from tutorials found on the Internet, few are able to quickly pick up C++ without studying a well-written...

Ixrec

2:06 PM

the norm is that up in the main rendering loop you have a try/catch for everything you had no way of "properly" handling and then show the user an error message instead of outright crashing

bigcodeszzer

Well, in the case of file opening

Instead of try catch, I just put if(file.open()){

Ixrec

just to save you some time, in many cases file opening works just fine with return codes

bigcodeszzer

I feel like there would be an equivalent for memory allocation

Ixrec

I'm not saying exceptions are always the right move for that the way they kind of are with memory allocation

bigcodeszzer

Is there a flag somewhere that you could test

Ixrec

2:08 PM

@bigcodeszzer there is if you're allocation and deallocating manually; iirc malloc() returns a null pointer if it fails

bigcodeszzer

I've never done that yet.

Ixrec

because you're using C++, not C

bigcodeszzer

Ah. That makes sense

BarryTheHatchet

Handling errors by wrapping everything in if and else is the philosophical cousin of putting all of your code in a single function

bigcodeszzer

I only use new though

BarryTheHatchet

2:08 PM

Don't use new

Ixrec

which is correct

BarryTheHatchet

no it's not

Ixrec

well, new is better than malloc

BarryTheHatchet

yes

Ixrec

but even new you shouldn't need anymore

BarryTheHatchet

2:09 PM

yep

bigcodeszzer

Yeah.

Ixrec

which is an entirely separate discussion

no

yeppers

na

2:09 PM

Vector and what not

BarryTheHatchet

I was thinking more of make_unique

Ixrec

I think smart pointers and "STL" containers are a very different subject from exceptions vs return codes

BarryTheHatchet

>.<

bigcodeszzer

That I don't know, but vector handling dynamic memory definately makes sense

assuming the person using it knows how to use it properly.

Ixrec

yes I've started saying "STL" in scare quotes just to annoy you

=)

BarryTheHatchet

2:10 PM

@Ixrec They're highly related because manual memory management produces largely exception-unsafe code

so if for some reason you must use new everywhere, perhaps you're better off with return codes

bigcodeszzer

I know what STL is

BarryTheHatchet

bet you don't

Ixrec

@bigcodeszzer that was directed at Barry

BarryTheHatchet

@Ixrec the quotes sanitise it ;p

Ixrec

Barry dislikes it when people use "STL" to mean things that aren't actually the STL, which everyone does

bigcodeszzer

2:10 PM

its the library that contains shit like vector

beyond that I don't care

BarryTheHatchet

well said

"shit like vector" wow

Ixrec

hey, std::vector is what separates us from the animals

bigcodeszzer

I had an assigment that supposedly was teaching stl algorithms. But it was very poorly written

BarryTheHatchet

where do you work, @bigcodeszzer?

bigcodeszzer

I'm in school

Ixrec

2:12 PM

so...let's get back to exceptions

BarryTheHatchet

@enderland um, no, that was about someone marching in and calling everyone (well, in this case, calling gnat) "rude" and "snobby". read more carefully before attacking me please. use your moderator brain!

bigcodeszzer

Separates us from the annimals is a bit of a stretch. Everyone should know /how/ to use raw pointers

Ixrec

yes all that stuff is relevant to exception safety but I think we have to convince @bigcodeszzer that exceptions have a point before talking about exception safety makes any sense

bigcodeszzer

It's ugly.

try catch throw is damn ugly.

BarryTheHatchet

@bigcodeszzer oh I wouldn't have thought that okay

Ixrec

2:13 PM

if you're using it for control flow it's ugly

bigcodeszzer

Exactly. ixrec

Ixrec

if you use error codes to handle all possible memory allocation failures that's also super ugly

BarryTheHatchet

try {
   this->conversation.have();
}
catch (std::logic_error&) {
   # Debugger currently breaking on this line
   exit(-1);
}

bigcodeszzer

Which is why, what I've gathered, is my prof is teaching it wrong.

Kit Z. Fox

@BarryTheHatchet You missed a curly bra--oh.

Ixrec

2:14 PM

I can imagine uni assignments fail to fully motivate exceptions as they tend to be completely unnecessary in 100-line programs

bigcodeszzer

Makes sense

BarryTheHatchet

those who can't, teach, etc

@Ixrec yeah that's true and tough

Ixrec

my uni made me write 1000-line programs where they did just barely have a point =D

BarryTheHatchet

exceptions, like a properly structured program design, may be the sort of thing that's hard to grok the purpose of until you actually write a full-scale, uncontrived program for the first time

bigcodeszzer

I've learned more dicking around in java on my own time

Ixrec

2:15 PM

many people report that to some degree

bigcodeszzer

But being forced to use c++ (Cause otherwise I Would never have learned it at all) definately made me better at doing things that would have otherwise never practiced

BarryTheHatchet

Would you please cut out the endless swearing?

It's BEEEEEEEP annoying.

bigcodeszzer

that's hardly swearing

hahaha

swearing is good for you

Ixrec

the uni stuff is valuable not for the actual languages you learn but for the computer science background stuff on language design and compiler implementations and algorithm analysis and whatnot

BarryTheHatchet

Not when my boss sees it on my screen, it isn't

bigcodeszzer

2:16 PM

Your boss probably doesn't know what he's talking about

Ixrec

@bigcodeszzer I find it's a cop-out when I'm too lazy to think of a properly expressive word that conveys my real feelings

bigcodeszzer

because the biz is designed to make money, not good code.

Ixrec

I'd rather pause for a second and call something "exceptionally harmful and unacceptable" than "f---ing stupid"

bigcodeszzer

That's a good point xrec, but limiting your vocabulary is better?

Ixrec

that's the exact opposite of what I was arguing lol

bigcodeszzer

2:17 PM

effing stupid adds the 'and annoying, and makes me frustrated, and I want to stab people who do said thing'

in one word!

BarryTheHatchet

@enderland Actually, on reading your message again, I'm very annoyed by it. I haven't "trolled" anyone. Actionless? Why would I start moderating a site whose moderation rules I disagree with? I've said that before, too. Though I do help you close and delete obviously bad questions. So what's your problem? Why do you feel the need to "come at me" like this? You're a moderator; you should know better. Please delete your remarks or at the least refrain from making them in the future. Thanks.

That's three times now. Next time I'll be forced to flag.

bigcodeszzer

Anyways, I think my understanding of exceptions is half complete.

Ixrec

so...exceptions

BarryTheHatchet

Exceptions seem to be the exception to this conversation

bigcodeszzer

I know how/where you should /not/ use them

But I don't know how/where you should use them.

Ixrec

2:19 PM

this conversation is a good example of what happens when you use error codes instead of exceptions; all the code is about propagating codes instead of doing anything useful

BarryTheHatchet

You will find out as you ~~grow in~~ start your career

Best not to judge them until that day comes

Ixrec

@bigcodeszzer is the question where a language should use them or when you should throw exceptions yourself?

bigcodeszzer

If you can't give me an example, I would almost make the stipulation you don't understand the topic properly.

Ixrec

we did...

bigcodeszzer

You ask me about something I know, I can explain it very simply, and very quickly. Probably with a swear word even.

BarryTheHatchet

2:20 PM

I gave you a great example

bigcodeszzer

Dynamic memory?

BarryTheHatchet

@bigcodeszzer And they wonder why I suspected you of trolling. :smh:

Ixrec

I've been sticking to just one example but mostly because it's so much more convincing than all the other examples

bigcodeszzer

I do troll, but not on programming forums.

programming is serious

BarryTheHatchet

Right well since this is not a programming forum, you're in the clear to troll away

bigcodeszzer

2:21 PM

whatever

BarryTheHatchet

I feel like we didn't do April Fools' properly in this room

Ixrec

you guys need to stop throwing so many checked ConversationExceptions so we can start ignoring them and staying on point

bigcodeszzer

okay so your example code was

BarryTheHatchet

there is no small example code that'll convince you that exceptions are needed

Ixrec

(no idea if @bigcodeszzer knows about checked exceptions)

BarryTheHatchet

2:21 PM

Ixrec explained that already

bigcodeszzer

try{ this->conversation.have() } ....

how is is your exception handling any different from an if() check

BarryTheHatchet

have fun @Ixrec

I'm going to do something more productive, like weeing all over the wall

bigcodeszzer

You say, you would have to have error codes everywhere, while it seems you would have to similiarly have try/catch blocks everywhere

Ixrec

@bigcodeszzer because the try can be placed higher up the call stack if we want, and the actual throw inside have() might be much farther down the call stack than it looks

obviously in the one-line psuedocode it's not, as previously mentioned

bigcodeszzer

if(file.isopen()){

lowerstackfunction()
}

Ixrec

2:23 PM

no, you only need one try/catch block instead of zillions of error code returns, that's the whole point

Ampt

If the code is going to fail and you have no reasonable way to handle it in that method, let it fail, and the caller can sort it out.

bigcodeszzer

Is that not what I wrote?

BarryTheHatchet

no

bigcodeszzer

You would rather have it fail in lowerstackfunction()

Ixrec

@bigcodeszzer if you mean me, I was responding to your claim that using exceptions means "try/catch blocks everywhere" which as I just said is not at all true

BarryTheHatchet

2:24 PM

no.

if file cannot be opened then it should already have thrown an exception. you don't even get to lowerstackfunction()

the caller can then handle that exception, potentially with the same code that handles other exceptions, some of which may come from within lowerstackfunction

bigcodeszzer

bool higherstackfucntion(ifstream myfile)

BarryTheHatchet

instead of needing thirty if statements, unclear object lifetime semantics, and a contaminated API

bigcodeszzer

return true/false

BarryTheHatchet

ok just do that then

bigcodeszzer

But they are the same?

BarryTheHatchet

2:26 PM

yep sure okay

bigcodeszzer

notice the question mark

notice the answer

then explain?

no thanks

	fooWithExceptions() {
		try {
			doWorkWithExceptions();
		} catch(WorkException e) {
			std::cout << "Failed to do work: " << e.what() << std::endl;
		}
	}

	doWorkWithExceptions() {
		a();
		b();
		c();
		d();
	}

	fooWithErrorCodes() {
		int rc = doWorkWithErrorCodes();
		if(rc != 0) {
			std::cout << "Failed to do work: rc=" << rc << std::endl;
		}
	}

	doWorkWithErrorCodes() {
		int rc = a();
		if(rc != 0) {
			return rc;
		}
		rc = b();
		if(rc != 0) {
			return rc;
		}
		rc = c();
		if(rc != 0) {

does that help at all?

in spite of SE chat hating indentation

BarryTheHatchet

2:27 PM

and that's the short version

assumes all your functions use the same error code set

bigcodeszzer

hold on

Ixrec

:28768936 huh, I thought that just added an indentation level, thanks

BarryTheHatchet

@Ixrec it does

that indentation level makes it code-markdown

Ixrec

oh I was missing indentation on the blank lines, right

BarryTheHatchet

newb

Kit Z. Fox

2:29 PM

@BarryTheHatchet *n00b

bigcodeszzer

Okay. I see what you're getting at

But why not:

Ixrec

@bigcodeszzer progress! \o/

BarryTheHatchet

Teal'c's interview

►

reminds me very much of this (~1:10)

bigcodeszzer

rc = a();
rc = b(rc);
rc = c(rc);

with rc as a bool

Ixrec

why would b want to know the return value of an unrelated function?

designing b() that way effectively makes it so b() can only ever be correctly called immediately after a()

bigcodeszzer

2:31 PM

Cause then you just put an at the top of all the functions

prolly just be one function, with different params each time

Um. Hold on.

Ixrec

I have no idea what those last two statements mean

BarryTheHatchet

"um" means "I have no idea what to say"

Ixrec

a few implicit assumptions I'm making in that example code:

BarryTheHatchet

"hold on" means "brb going to find out what to say"

Ixrec

1) a, b, c, d have absolutely nothing to do with each other in principle

bigcodeszzer

2:32 PM

Well, what is wrong with the error codes example?

It's ugly, yes? Too many if() calls?

Ixrec

2) continuing program execution after a, b, c, or d has failed is undefined behavior

bigcodeszzer

Well, I assumed you wanted it to stop

Ixrec

@bigcodeszzer half of the code in doWorkWithErrorCodes() is taken up by properly propagating the error codes, instead of doing real work

that's the problem

BarryTheHatchet

a bit like this room

Ixrec

now imagine if these were functions that actually acquired resources which later have to be freed

BarryTheHatchet

2:34 PM

half of the day in the Whiteboard is taken up by properly trolling each other, instead of doing real work

not sure it's a problem though

bigcodeszzer

Well, I'm suggesting that with a little re-organization, namely, functional paradigm style, it would be shorter

Ixrec

in C++ with exceptions you get the RAII guarantee that exceptions unwind the stack and call everybody's destructors

BarryTheHatchet

actually that's not quite true

it's only so if you catch the exception somewhere

bigcodeszzer

I feel like if you need to be calling the destructors the scopes are managed poorly.

Ixrec

true, but we're assuming the goal is not to outright crash the program on these failures

bigcodeszzer

2:35 PM

Well I would have if() at the front of each function

Ixrec

@bigcodeszzer you don't need to call them, that's something C++ does for you

bigcodeszzer

firstfunction(rc){

if(!rc){
return false

else{

...

}

BarryTheHatchet

@bigcodeszzer no, it's the opposite. having destructors is the very definition of managing scopes properly

@bigcodeszzer that violates SRP and introduces a massive coupling between your functions

Ixrec

that's the whole point of new, delete, constructors, and destructors, almost all allocation/deallocation code gets called for you by the language when it should be called so you don't have to waste time figuring it out

bigcodeszzer

Yes.

Ixrec

2:36 PM

5 mins ago, by Ixrec

designing b() that way effectively makes it so b() can only ever be correctly called immediately after a()

bigcodeszzer

Then why does RAII need to do that?

Oh

Ixrec

I don't understand the question

BarryTheHatchet

neither do I

bigcodeszzer

nvm its not important

BarryTheHatchet

but you're so close to a breakthrough

Ixrec

2:37 PM

the whole point of objects having destructors is so that their cleanup code gets called automatically when needed, so you don't have to keep track of when you need to call it

bigcodeszzer

Yes. Which happens when they go out of scope

You said: in C++ with exceptions you get the RAII guarantee that exceptions unwind the stack and call everybody's destructors

What does that mean?

Destructors are supposed to be automatic, no?

Ixrec

that means that everybody's destructors get called whether the function exits normally or whether the function exits prematurely due to an exception getting thrown

roughly speaking don't nitpick that please

bigcodeszzer

Which leads me back to stipulation a) something is wrong with your code

Ixrec

why?

bigcodeszzer

Well I would use a temporary variable

BarryTheHatchet

2:40 PM

wtf

bigcodeszzer

and put an if somewher

and if that if doesn't pass, I don't insert the dynamic memory

Ixrec

"insert the dynamic memory"?

bigcodeszzer

It gets deleted immediately.

vector.push_back(tempvariable)

BarryTheHatchet

that's what your destructor should be doing

bigcodeszzer

Well yeah

BarryTheHatchet

2:41 PM

why are you trying to make life harder for yourself

Ixrec

doWorkWithErrorCodes() {
	int rc = a();
	if(rc != 0) {
		freeAllResourcesCreatedByA();
		return rc;
	}
	rc = b();
	if(rc != 0) {
		freeAllResourcesCreatedByB();
		freeAllResourcesCreatedByA();
		return rc;
	}
	rc = c();
	if(rc != 0) {
		freeAllResourcesCreatedByC();
		freeAllResourcesCreatedByB();
		freeAllResourcesCreatedByA();
		return rc;
	}
	rc = d();
	freeAllResourcesCreatedByD();
	freeAllResourcesCreatedByC();
	freeAllResourcesCreatedByB();
	freeAllResourcesCreatedByA();
	return rc;
}

2

bigcodeszzer

fn_allocate(){

sometype tempvariable = new sometype();

if(condition){

somevector.push_back(tempvar)

BarryTheHatchet

You mean sometype*

Aaron Hall

:)

Kit Z. Fox

@BarryTheHatchet grabs wrist and forces hand into face why are you punching yourself? ;)

bigcodeszzer

2:42 PM

Uh. Probably

BarryTheHatchet

and if you'd written auto tempvariable = std::make_unique<sometype>() instead then you wouldn't need to worry about that variable any more, whether you get an exception or not. as it is now, you're leaking that memory because you didn't delete it. so even with this trivial example, you've already got a terrible bug caused by your most-effort approach.

Ixrec

so you're just letting the memory leak?

BarryTheHatchet

also, why are you creating an object outside of the conditional if you only need it inside the conditional?

Kit Z. Fox

At least put some towels down.

Ixrec

as Barry said, using RAII classes like smart pointers ensures everything would get cleaned up properly without you even having to think about these things

bigcodeszzer

2:43 PM

Hmmm

Well what I'm saying is, the sometype would be an object.

Ixrec

if you new it, it's a pointer

bigcodeszzer

And therefore, if it doesn't get push_back, the function just ends

BarryTheHatchet

well no the sometype is still an object

bigcodeszzer

and then it goes out of scope

BarryTheHatchet

mind you, the pointer is an object too :)

bigcodeszzer

2:44 PM

and calls destructor

BarryTheHatchet

@bigcodeszzer no it doesn't. you dynamically allocated it

Ixrec

raw pointers don't have destructors

that's the whole reason smart pointers exist

BarryTheHatchet

it's like you're taking all the benefits of doing things our way, and attributing them to doing things your way instead

bigcodeszzer

If I allocate something with new.

Ixrec

if you do type* = new ... and never call delete on that pointer later, you've leaked whatever it points to

bigcodeszzer

2:45 PM

It doesn't get destroyed when the pointer goes out of scope?

BarryTheHatchet

no...

Ixrec

59 secs ago, by Ixrec

raw pointers don't have destructors

BarryTheHatchet

otherwise your vector would contain a dangling pointer wouldn't it

bigcodeszzer

What if I don't use new?

BarryTheHatchet

then it will

that's RAII

bigcodeszzer

2:46 PM

So object = object();

BarryTheHatchet

that's silly

Ixrec

or if you use things like std::make_unique that give you a proper smart pointer

bigcodeszzer

That would work?

Ixrec

or just object();

BarryTheHatchet

just type object; please

yes

Ixrec

2:46 PM

or object;

BarryTheHatchet

@Ixrec that's not valid

bigcodeszzer

Okay, well barring that mistake.

Ixrec

the extra copy whatever is unnecessary

@BarryTheHatchet Object object; is not valid?

BarryTheHatchet

@Ixrec it's not copy-assignment. it's copy-initialisation.

@Ixrec that's not what you wrote :P

Ixrec

you were omitting the Object part too

BarryTheHatchet

2:47 PM

@Ixrec no I wasn't

although I called it type

Ixrec

ok never mind conversation moving too fast

bigcodeszzer

fn_allocate(){

sometype tempvariable = sometype();

if(condition){

somevector.push_back(tempvar)

}

}

BarryTheHatchet

whooshhhhhhh

bigcodeszzer

yeah

BarryTheHatchet

I'm in desperate need of some calming music

bigcodeszzer

2:48 PM

Listen to tom waits

It will also make you more insane

Kit Z. Fox

Small Change got rained on with his own .38.

BarryTheHatchet

what?

Ixrec

admittedly, destructors get called whether you return error codes or throw exceptions

bigcodeszzer

30 aught 6

Kit Z. Fox

No. It's .38.

bigcodeszzer

2:48 PM

okay

so have we trumped the dynamic allocation issue?

Ixrec

but a failure inside a constructor is a classic example of an error where exceptions really are your only option

BarryTheHatchet

trumped?

Aaron Hall

18 hours ago, by Aaron Hall

♪ "Barry in charge, of our days, and our nights. (Barry in charge) Barry in charge, of our wrongs, and our rights..." ♪

bigcodeszzer

I might argue that you shouldn't be allocating in a constructor

Ixrec

what do you think the std::vector constructor does?

BarryTheHatchet

2:49 PM

Stargate Universe Soundtrack - Countdown To Destiny (Joel Goldsmith)

►

bigcodeszzer

Well, that's them

BarryTheHatchet

there we go

@bigcodeszzer it's best practice. it's literally the appropriate place for allocations

Kit Z. Fox

I used to date a guy who looked like Tom Waits.

bigcodeszzer

and they probably used a try catch I guess, to make that work

BarryTheHatchet

doubtful

Ixrec

2:50 PM

most classes shouldn't allocate memory directly only because standard containers and smart pointers can be used for that stuff

BarryTheHatchet

you're supposed to handle the exception further down the call stack

bigcodeszzer

what if you call a function in the constructor?

Kit Z. Fox

Tom Waits - Small Change

that handles that

then what

2:50 PM

I've completely lost you again

bigcodeszzer

Then returns an error code

BarryTheHatchet

what can you do with an object that failed to initialise

Ixrec

if you try to construct a std::vector and there's no memory left to allocate, what alternative is there to throwing a std::bad_alloc?

bigcodeszzer

I don't know.

There must be a fail flag or something

to test

Ixrec

on what?

Aaron Hall

2:51 PM

He should learn Python, that's easier.

BarryTheHatchet

dude. he's trolling. I'm now 80% sure of it.

Ixrec

there is no std::vector, it couldn't be constructed

BarryTheHatchet

you've had the same discussion four times

Ixrec

this isn't like C where you allocate a blob of memory and then build an object in that memory as a separate step

bigcodeszzer

Okay. Situation is, you called a constructor, the constructor couldn't allocate the memory

Ixrec

2:52 PM

the whole idea of new was to combine those steps into one

bigcodeszzer

a vector lets say

Ixrec

mm-hmm

Aaron Hall

Someone should star that comment. It's pure gold.

Ixrec

which one?

^ applies to half of what's been said in this convo

Aaron Hall

Mine, about Python.

:D

BarryTheHatchet

2:53 PM

ssh Aaron

bigcodeszzer

Weak

Anyways.

So you have

try{

new object()

}catch{

...
}

shoot me

lol

bang

yay guaranteed memory leaks

Aaron Hall

2:55 PM

do you call that insignifigant white-space?

whatsisname

im here now

BarryTheHatchet

it's not even valid syntax

Ixrec

right on time

BarryTheHatchet

@whatsisname RUN AWAY REALLY QUICKLY

bigcodeszzer

That's hot

whatsisname

2:55 PM

rly?

BarryTheHatchet

rly.

Ixrec

@whatsisname unless you're here for the schadenfreude

Aaron Hall

Hey, I got Spacemacs installed on my host/laptop last night at the emacs hack night.

whatsisname

not really

i just saw there was a chat notification for me

Ixrec

considering none of us know what the convo is supposed to be about it's probably a good idea to ignore it

bigcodeszzer

2:57 PM

whatsisname you said you could demonstrate a case where try/catch would be a better option than error codes

ixrec and barry are saying that the case is initializing with a constructor inside a try/catch block?

whatsisname

there's a ton of examples

bigcodeszzer

and catching out of memory exception

Well the main point, as I see it, is that try/catch shouldn't be used in flow control.

DeliriousSyntax

This account is scheduled to be deleted in 23 hours. To cancel deletion, click here.

IMPORTANT: If you've requested deletion on other Stack Exchange sites, you'll need to cancel those separately by visiting each site's profile.

bigcodeszzer

Which was my main issue.

Ixrec

we all agree on that

none of our examples have been advocating exceptions as flow control

whatsisname

2:59 PM

now, before i even bother, is it even possible to convince you of their value?

because obviously anything with exceptions can be implemented without them, because at the end of the day every boils down to conditional jumps

bigcodeszzer

Of course.

thank you