Coding Horrors: You can gather all your technical debt into one easy framework!

The Something Awful Forums > Discussion > Serious Hardware/Software Crap > The Cavern of COBOL > Coding Horrors: You can gather all your technical debt into one easy framework!

«‹›1503 »

Volguus: Mar 3, 2009

Hughlander posted:

For an extremely narrow set of problems that actually is really loving cool.

That No-op garbage collector is actually envisioned to be the skeleton upon which other collectors can be written. Yes, you're right that in a very specific set of circumstances it itself could be useful as is, but most of the time it'll be the base of other garbage collectors. That's the idea behind it, as I understand it.

# ? Feb 20, 2019 04:06

Adbot: ADBOT LOVES YOU

# ? May 17, 2024 19:39

senrath: Nov 4, 2009; Look Professor, a destruct switch!

Volguus posted:

That No-op garbage collector is actually envisioned to be the skeleton upon which other collectors can be written. Yes, you're right that in a very specific set of circumstances it itself could be useful as is, but most of the time it'll be the base of other garbage collectors. That's the idea behind it, as I understand it.

As far as I can tell from the link the no-op gc was designed to be used as is, not as a skeleton for further gcs to be written with. All of the things listed under "Motivations" are real (if niche) use cases, and none of them are "provide a framework for a functional GC to be built on."

# ? Feb 20, 2019 05:01

Carbon dioxide: Oct 9, 2012

https://stackoverflow.com/a/39150189

# ? Feb 20, 2019 17:44

Joda: Apr 24, 2010; When I'm off, I just like to really let go and have fun, y'know?; Fun Shoe

Carbon dioxide posted:

https://stackoverflow.com/a/39150189

quote:

straightforward

How convoluted must the rest of the language be, if that constitutes as a straight-forward way to do a basic datetime operation.

# ? Feb 20, 2019 17:56

Munkeymon: Aug 14, 2003; Motherfucker's got an
armor-piercing crowbar! Rigoddamndicu𝜆ous.

Carbon dioxide posted:

https://stackoverflow.com/a/39150189

Which one of you is this?

quote:

Thanks. This is disgusting. I love it. � wchargin

# ? Feb 20, 2019 18:08

Suspicious Dish: Sep 24, 2011; 2020 is the year of linux on the desktop, bro; Fun Shoe

ISO 8601 is far from straightforward

# ? Feb 20, 2019 18:08

Volguus: Mar 3, 2009

senrath posted:

As far as I can tell from the link the no-op gc was designed to be used as is, not as a skeleton for further gcs to be written with. All of the things listed under "Motivations" are real (if niche) use cases, and none of them are "provide a framework for a functional GC to be built on."

Here's where i've read about it: https://blog.idrsolutions.com/2018/04/java-10-improvements-to-garbage-collection-explained-in-5-minutes/

You're right that it is not a stated "motivation" of the GC, but it can obviously be used as such. Allocating memory is a (relatively) solved issue, not much to do there. De-allocating it is where the challenge lies.

# ? Feb 20, 2019 18:34

Carbon dioxide: Oct 9, 2012

If you scroll up you can see this kind of trickery is no longer necessary since Python 3.7, but yeah, took 'em long enough.

# ? Feb 20, 2019 18:50

Plorkyeran: Mar 22, 2007; To Escape The Shackles Of The Old Forums, We Must Reject The Tribal Negativity He Endorsed

If Java was more suitable for writing CLI programs then the no-op GC would probably be a pretty commonly used thing. It's a really easy way to pick up a lot of performance when the OS is going to GC your whole program before it matters anyway.

# ? Feb 20, 2019 18:53

JawnV6: Jul 4, 2004; So hot ...

Carbon dioxide posted:

https://stackoverflow.com/a/39150189

Wasn't python's whole deal looking down smugly on perl for having more than one way to do something?

# ? Feb 20, 2019 19:11

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

JawnV6 posted:

Wasn't python's whole deal looking down smugly on perl for having more than one way to do something?

yes, and having everything you�d need built into the standard library

then they decided instead that they would build in thirty different ways to format strings, while recommending using third party libraries instead of the stdlib for important things because it�s easier to fix bugs if you can release things whenever instead of waiting for an update to the whole of python

so ... yeah, most of the early python philosophy has been dead for a long time

# ? Feb 20, 2019 19:19

ToxicFrog: Apr 26, 2008

Plorkyeran posted:

If Java was more suitable for writing CLI programs then the no-op GC would probably be a pretty commonly used thing. It's a really easy way to pick up a lot of performance when the OS is going to GC your whole program before it matters anyway.

That's one of the motivations they mention, in fact!

Joda posted:

How convoluted must the rest of the language be, if that constitutes as a straight-forward way to do a basic datetime operation.

It's not so much convolution (in the language, anyways; anything involving dates and times is intrisically convoluted) as a hole in the builtin datetime library -- specifically, while it will happily generate ISO8601 dates, it can only parse a subset of them, since ISO8601, it turns out, offers some leeway in how formatting works. In particular, the time zone at the end can be any of:

code:

""
"Z"
"�HH"
"�HH:MM"
"�HHMM"

and strptime() only supports the first and last of those.

Making matters worse, the OP is asking specifically about RFC3339, a restricted subset of ISO8601 which only permits the middle three, i.e. exactly the ones Python's datetime parser doesn't support; and to add to the confusion, a number of the people replying have misread "RFC3339" as "ISO8601" and are answering the wrong question.

So at this point, the question basically reduces to "given that the language doesn't have a built in parser for this specific date format, what's your favourite 3rd-party date parsing library and/or favourite roll-your-own date parser" and that's pretty much always going to be a shitshow.

(FWIW, my answer would be "use a recent python that supports this format", and if that's not possible, "smoosh the RFC3339 date through a pair of regexes to turn it into an ISO8601 date strptime() can understand and then use that".)

# ? Feb 20, 2019 19:33

Loezi: Dec 18, 2012; Never buy the cheap stuff

So Nike just bricked a bunch of shoes with a bad firmware update process.

https://twitter.com/Jaxbot/status/1097631614153175041

# ? Feb 20, 2019 19:37

Rubellavator: Aug 16, 2007

Loezi posted:

So Nike just bricked a bunch of shoes with a bad firmware update process.

This sentence makes me feel old. Darn kids and their cyber shoes

# ? Feb 20, 2019 19:41

Thermopyle: Jul 1, 2003; ...the stupid are cocksure while the intelligent are full of doubt. �Bertrand Russell

Carbon dioxide posted:

If you scroll up you can see this kind of trickery is no longer necessary since Python 3.7, but yeah, took 'em long enough.

Only for a subset of valid strings. AKA, strings generated with datetime.isoformat.

# ? Feb 20, 2019 21:10

Scaramouche: Mar 26, 2001; SPACE FACE! SPACE FACE!

The secret to date time management is to do it one way and never ever change it. And if you're accepting input for it reject anything that even smells funny.

All that said I'm still reeling over the Solidity stuff.

# ? Feb 20, 2019 21:26

Scaramouche: Mar 26, 2001; SPACE FACE! SPACE FACE!

It's get better, Solidity has actual defenders on /ethereum
https://www.reddit.com/r/ethereum/comments/72096k/is_solidity_a_great_language_what_does_the/

# ? Feb 20, 2019 22:33

Mad Jaqk: Jun 2, 2013

Rubellavator posted:

This sentence makes me feel old. Darn kids and their cyber shoes

Calling them "cyber shoes" was enough to make me interested instead of angry at the entire concept.

# ? Feb 20, 2019 22:43

QuarkJets: Sep 8, 2008

What motivates a person to desire cybershoes?

# ? Feb 20, 2019 22:58

Adhemar: Jan 21, 2004; Kellner, da ist ein scheussliches Biest in meiner Suppe.

Plorkyeran posted:

If Java was more suitable for writing CLI programs then the no-op GC would probably be a pretty commonly used thing. It's a really easy way to pick up a lot of performance when the OS is going to GC your whole program before it matters anyway.

Exactly. Disabling GC is also very common in HFT (high frequency trading) and has been for years. Pausing for GC is a dealbreaker there, so why bother if your trading app only has to run until the market closes? RAM is cheap, so just install enough to last the trading day.

# ? Feb 20, 2019 23:23

xtal: Jan 9, 2011; by Fluffdaddy

I know that people turn the GC off with some PHP and Ruby programs as well. With Ruby on Rails some people disable GC during the duration of a web request.

# ? Feb 20, 2019 23:40

Factor Mystic: Mar 20, 2006; Baby's First Post-Apocalyptic Fiction

QuarkJets posted:

What motivates a person to desire cybershoes?

only cybershoes look good when I'm wearing my cyberpants & cyberjacket

# ? Feb 21, 2019 02:18

Dylan16807: May 12, 2010

ToxicFrog posted:

It's not so much convolution (in the language, anyways; anything involving dates and times is intrisically convoluted) as a hole in the builtin datetime library -- specifically, while it will happily generate ISO8601 dates, it can only parse a subset of them, since ISO8601, it turns out, offers some leeway in how formatting works. In particular, the time zone at the end can be any of:
code:
""
"Z"
"�HH"
"�HH:MM"
"�HHMM"
and strptime() only supports the first and last of those.

Making matters worse, the OP is asking specifically about RFC3339, a restricted subset of ISO8601 which only permits the middle three, i.e. exactly the ones Python's datetime parser doesn't support; and to add to the confusion, a number of the people replying have misread "RFC3339" as "ISO8601" and are answering the wrong question.

So at this point, the question basically reduces to "given that the language doesn't have a built in parser for this specific date format, what's your favourite 3rd-party date parsing library and/or favourite roll-your-own date parser" and that's pretty much always going to be a shitshow.

(FWIW, my answer would be "use a recent python that supports this format", and if that's not possible, "smoosh the RFC3339 date through a pair of regexes to turn it into an ISO8601 date strptime() can understand and then use that".)

Does any python version have a parser that supports "Z"? It seems like even with fromisoformat you need to throw in a .replace("Z", "+00:00")

Dylan16807 fucked around with this message at 02:52 on Feb 21, 2019

# ? Feb 21, 2019 02:48

senrath: Nov 4, 2009; Look Professor, a destruct switch!

QuarkJets posted:

What motivates a person to desire cybershoes?

Theoretically they allow a better fit than traditional laces, but 99.99999% of the people who buy these things will never be in a position where that will actually be relevant.

# ? Feb 21, 2019 09:28

RPATDO_LAMD: Mar 22, 2013; 🐘🪠🍆

Bruegels Fuckbooks posted:

i do think a language with no garbage collector which also doesn't support manual memory management is complete genius and i wish all languages were like that. "gently caress it, nothing we can do about that memory leak."

Pretty sure Rust does this. You have to try pretty hard to slip a memory leak by the compiler's checks though.

# ? Feb 21, 2019 09:51

Joda: Apr 24, 2010; When I'm off, I just like to really let go and have fun, y'know?; Fun Shoe

SQL code:

 
IF @Value > 1
    SET @Value = @Value

Found this gem at work. No elses or anything either.

# ? Feb 21, 2019 12:26

redleader: Aug 18, 2005; Engage according to operational parameters

Scaramouche posted:

It's get better, Solidity has actual defenders on /ethereum
https://www.reddit.com/r/ethereum/comments/72096k/is_solidity_a_great_language_what_does_the/

quote:

Of course we all might wish that the Solidity designers would have deeply studied preexisting formalisms for smart contracts

Why study pre-existing ideas about smart contacts for a language designed for smart contracts, when you can instead just slap some poo poo together and throw it at a blockchain?

Those defenders :psyduck:

# ? Feb 21, 2019 12:34

Spatial: Nov 15, 2007

"The Unstoppable Will of Code"

# ? Feb 21, 2019 12:36

Volguus: Mar 3, 2009

redleader posted:

Why study pre-existing ideas about smart contacts for a language designed for smart contracts, when you can instead just slap some poo poo together and throw it at a blockchain?

Those defenders

Well, those defenders are right that the creators of Ethereum needed something right then and there not 10 years later. It just neededn't be Solidity.

# ? Feb 21, 2019 13:03

Nude: Nov 16, 2014; I have no idea what I'm doing.

QuarkJets posted:

What motivates a person to desire cybershoes?

From what I read the main pitch is you don't have to tie your shoes anymore...

You use an app instead.

# ? Feb 21, 2019 13:59

Bruegels Fuckbooks: Sep 14, 2004; Now, listen - I know the two of you are very different from each other in a lot of ways, but you have to understand that as far as Grandpa's concerned, you're both pieces of shit! Yeah. I can prove it mathematically.

RPATDO_LAMD posted:

Pretty sure Rust does this. You have to try pretty hard to slip a memory leak by the compiler's checks though.

Releasing the object when it's no longer referenced/goes out of scope is like, memory management (rust's checks etc. ensure memory is handled like c++ smart pointers). Solidity has no mechanism for the programmer to reclaim memory that has been previously used.

# ? Feb 21, 2019 14:41

iospace: Jan 19, 2038

Joda posted:

SQL code:
 
IF @Value > 1
    SET @Value = @Value
Found this gem at work. No elses or anything either.

Did you find who wrote that and ask them about it?

# ? Feb 21, 2019 15:04

Joda: Apr 24, 2010; When I'm off, I just like to really let go and have fun, y'know?; Fun Shoe

iospace posted:

Did you find who wrote that and ask them about it?

No it's too old to ask about, based on context it was definitely meant to clamp the value.

# ? Feb 21, 2019 15:16

Cuntpunch: Oct 3, 2003; A monkey in a long line of kings

Joda posted:

SQL code:
 
IF @Value > 1
    SET @Value = @Value
Found this gem at work. No elses or anything either.

code:

//C#
public class Foo
{
    private string _text;
   
    public Foo(string text)
    {
        if(text != null)
        {
            _text = text;
        }
    }
}

This 'pattern' was *rampant* at work and every time I asked the people doing it *why* they did it, they cited some sense of 'because we have always done it this way'. Some of the more junior folks, when asked *what* it does spent a lot of time and effort talking themselves in circles about how it was preventing null references.

# ? Feb 21, 2019 16:03

Munkeymon: Aug 14, 2003; Motherfucker's got an
armor-piercing crowbar! Rigoddamndicu𝜆ous.

QuarkJets posted:

What motivates a person to desire cybershoes?

A miserable pile of disposable income.

redleader posted:

Why study pre-existing ideas about smart contacts for a language designed for smart contracts, when you can instead just slap some poo poo together and throw it at a blockchain?

Those defenders

IIRC their first try was a pure functional language but people found it too hard to understand and use lol

# ? Feb 21, 2019 17:59

itskage: Aug 26, 2003

Cuntpunch posted:

code:
//C#
public class Foo
{
    private string _text;
   
    public Foo(string text)
    {
        if(text != null)
        {
            _text = text;
        }
    }
}
This 'pattern' was *rampant* at work and every time I asked the people doing it *why* they did it, they cited some sense of 'because we have always done it this way'. Some of the more junior folks, when asked *what* it does spent a lot of time and effort talking themselves in circles about how it was preventing null references.

https://docs.microsoft.com/en-us/dynamicsax-2012/developer/best-practices-for-accessor-methods

code:


container parmCode(container _code = conNull())

{

if (!prmIsDefault(_code))

{

code = _code;

}

return code;

}

e: phone posting so I'm not fixing that indentation.

# ? Feb 21, 2019 19:57

Hammerite: Mar 9, 2007; And you don't remember what I said here, either, but it was pompous and stupid.; Jade Ear Joe

itskage posted:

https://docs.microsoft.com/en-us/dynamicsax-2012/developer/best-practices-for-accessor-methods
code:
container parmCode(container _code = conNull())

{

if (!prmIsDefault(_code))

{

code = _code;

}

return code;

}
e: phone posting so I'm not fixing that indentation.

I'd never heard of the product this relates to, though it is apparently a .NET framework language. Do you really think it's probable that Cuntpunch's coworkers got it from there? At the risk of generalising from my own experience, I assume that most C# programmers also haven't heard of this thing. It has fewer than 300 questions on Stack Overflow, which surely is hardly anything.

# ? Feb 21, 2019 20:10

Athas: Aug 6, 2007; fuck that joker

Munkeymon posted:

IIRC their first try was a pure functional language but people found it too hard to understand and use lol

A pure functional language still with stupid type rules would not have made much of a difference. It doesn't look like the main problem with Solidity is how it handles effects.

# ? Feb 22, 2019 16:34

Munkeymon: Aug 14, 2003; Motherfucker's got an
armor-piercing crowbar! Rigoddamndicu𝜆ous.

Athas posted:

A pure functional language still with stupid type rules would not have made much of a difference. It doesn't look like the main problem with Solidity is how it handles effects.

Pure functional languages seem less prone to dumb as poo poo bugs than half-assed JS clones like Solidity, though. "If it compiles, it probably works" is something that gets said a lot about Haskell or Elm and that would be a good feature of an immutable codebase.

# ? Feb 22, 2019 16:40

Adbot: ADBOT LOVES YOU

# ? May 17, 2024 19:39

Athas: Aug 6, 2007; fuck that joker

Munkeymon posted:

Pure functional languages seem less prone to dumb as poo poo bugs than half-assed JS clones like Solidity, though. "If it compiles, it probably works" is something that gets said a lot about Haskell or Elm and that would be a good feature of an immutable codebase.

Really? I can come up with a ton of ideas for a pure functional language that I expect from the kind of mind that would create Solidity:

Dynamically typed, obviously.
All numbers are 512-bit integers.
Any value can be used as a function, which is string prefixing if it is string-like, or multiplication if it is number-like. So you can write 'x y'!
Implicit casting between strings and numbers, of course. Any string containing solely hexadecimal digits (ignoring whitespace) is interpreted as a hex number. Leading zeroes make it octal.
An expression using unbound names is implicitly considered a lambda taking parameters matching those unbound names. For example if 'x' and 'y' does not exist, the expression 'x+y' corresponds to Haskell's '\x y -> x + y'. The parameters are sorted alphabetically, so 'y+x' is also '\x y -> x + y'.
There is a record system, and any value can be extended with fields (Javascript-style) with the syntax 'object.fieldname = fieldvalue'. But it returns a copy of object, so purely functional, see?
If you try to read an unknown field, you get back numeric zero.
If you try to perform a mathematical operation on a function (e.g. 'f + 2'), then that creates an implicit lambda that applies the operation to the result of calling the function, i.e., this example becomes what you'd write in Haskell as '\x -> f x + 2'. This applies recursively if 'f x' returns a function, so be careful you don't blow the stack!
Speaking of the stack, the compiler does not perform tail-call optimisation, but recursion is still the only way to loop.

# ? Feb 22, 2019 17:11

The Something Awful Forums > Discussion > Serious Hardware/Software Crap > The Cavern of COBOL > Coding Horrors: You can gather all your technical debt into one easy framework!

«‹›1503 »