mcc
You are in a C/C++ file. You declare a (C/POD) static variable at the :: level (not inside of a function. Because it is static, under the C standard you know that it will be auto-initialized to the type's 0 value. And in the semantics of your program, this variable *should* be initialized to zero. So you:
Provide no initializer, since it defaults 0
8.6%
Type "= 0" or "= false" anyway, for readability
82.1%
Andi what the fuck are you talking about
9.3%
Poll ended at .
Oct 10, 2024 at 3:07amWeb
Show threadmccOct 10, 2024
Note: I acknowledge that this question is invalid, due to unclosed parenthesis on line 1:70
Show threadBrazMogu (i.e. Bruno Guedes)Oct 10, 2024
@mcc Thank you, I thought my parser was glitching
Show threadAllan ChowOct 10, 2024
@mcc I read this in strongbad's voice.
Show threadLuci RoseOct 10, 2024
@mcc the initializer also makes the code more resilient to refactoring into context where default 0 isn't standard
Show threadMichael WymanOct 10, 2024
@luci @mcc exactly this. Code copying leads to so many bugs due to no-longer-correct assumptions.
Show threadGarrett FleenorOct 10, 2024
@mcc always make your intent clear so the compiler has a chance to optimize. Also prevents compiler bugs from surprising you.
Show threadBradOct 10, 2024
@RoyAwesome @mcc Give the linker a chance to pessimize, fill the rdata segment with hardcoded zeroes. šŸ˜‰
Show threadrosalyn Oct 10, 2024

@mcc Ooh, I love interactive fiction!

> provide no initializer

You have been eaten by a grue.

Show threadPaul KhuongOct 10, 2024
@mcc Do you care about padding bits?
Show threadGreg TitusOct 10, 2024
@mcc "under the C standard you know" ... do I?!? I'm not certain I do. I'm pretty sure I have doubts and so I'm typing "= 0" not because of high-falutin readability concerns but because of fear of the unknown.
Show threadmccOct 10, 2024
@gregtitus this is an extremely fair point, but for purposes of this question you are role-playing a person who has recently checked the relevant c/c++ standard and verified this to be the case. (however we may still count your reply as a "type it for readability" because even if you have checked the standard and know this, you must assume your coworkers/collaborators have not)
Show threadfabiƔn caƱasOct 10, 2024
@mcc @gregtitus Role playing someone who read the standard, I add it anyway because I still don't know what the compiler will do. I can check today, but what about tomorrow?
Show threadBradOct 10, 2024
@mcc I once worked on embedded firmware that had a bug where it didnā€™t initialize BSS. Until this was fixed, every global variable had an initializer. šŸ˜…
Show threadmccOct 10, 2024
@bk1e :squeaks: noā€¦ no!
Show thread银åƼēŗæOct 10, 2024
@mcc I live in the auto-initialized variable. The value shall be initialized to me
Show threadpeter hessler @openbsdOct 10, 2024
@mcc is this one of those horrifying situations where C++ does something different from C?
Show threadmccOct 10, 2024
@phessler No, not this time. Although obviously [C++ types with initializers in C++] may behave different from [C types in C++].
Show threadRachel GreenhamOct 10, 2024
@mcc explicitly set. in fact I do exactly that in Java, if the initial value is significant. (ie: isnā€™t certain to be set elsewhere before it would be used.) ditto instance vars.
Show threadTom ForsythOct 10, 2024
@mcc The correct response is to panic and initialise it in a function anyway. Because anybody who believes compiler/language rules in C++ is cruising for a bruising.
Show threadOblomovOct 10, 2024
@TomF @mcc going to be a bit messy when they are static const
Show threadTom ForsythOct 10, 2024
@oblomov @mcc Those don't have storage.
Show threadOblomovOct 10, 2024

@TomF @mcc

doh, I should stop replying to tech threads at 7am

Show threadPeter SommerladOct 11, 2024
@TomF @oblomov @mcc
unless their address is taken.
Show threadOblomovOct 10, 2024
@mcc depends. If everything is going to be zero, I'll just trust the compiler to follow the standard. If other nearby values need to be initialized to something different than 0, I'll do the explicit thing for the null values.
Show threadM. VerdoneOct 10, 2024
@mcc I type as little as possible and wait for the compiler error to tell me what I should do.
Show threadfabiosantoscodeOct 10, 2024
@mcc I think someone could create C++ with the benefit of hindsight and make it a great language. There's probably library shelves worth of writing about why and how C++ is flawed, who could ask for better user research?
Show threadmccOct 10, 2024
@fabiosantoscode that might just be C#. Whether it's great can be a matter of opinion.
Show threadfabiosantoscodeOct 10, 2024
@mcc I find it way more tolerable than C at least
Show threadmccOct 10, 2024
@fabiosantoscode I would describe C# as "Pretty good!"
Show threadHenriOct 10, 2024
@mcc static outside of a function means something completely different and is just a global variable that is not linkable. And I would use `= {};` in C++ and C23
Show threadjenOct 11, 2024
@mcc Definitely explicit initialization in most languages. My personal exception is go, which has super consistent zero value behavior for all types and all declarations, which I acknowledge is complete vibes-based
Show threaddemizeOct 11, 2024

@mcc C spec concerns aside, I just dislike when people do things like that anyway, because if Iā€™m reading or reviewing the code it makes me have to think about their intent

code is language and language is about communication; if you donā€™t communicate your intent, you arenā€™t effectively using it, even if you think itā€™s more ā€œelegantā€

(note I would also take a comment, if itā€™s some sort of situation where the default value can change and you want this to always be the default, because ā€œ// initializes to the default value (currently 0)ā€ also communicates intent well)