Zig: software should be perfect [video] • Hacker News Board

Re: Zig: software should be perfect [video]

#51

Post by bjourne » Sun, Nov 11, 2018, 1:57 AM UTC

electrograv wrote: post #46

mschwaig wrote: post #23

The definition of 'perfect software' used in the talk is 'it gives you the correct output for every input in the input domain'. To that end no funny business like hidden allocations or hidden control flow should happen behind your back, because an out of memory error or some exception your code does not deal with explicitly would not be a correct output according to that definition. Of course you do not need that lev…

While I agree with the content of your post literally, I think we often underestimate the importance of software reliability and performance, and end up giving it less attention than it deserves. I understand the place for rapid prototyping etc., and that not every software application deals with life-and-death situations — but even those that aren’t, I think our industry suffers a bit here. For example, to this day…

It is entirely possible that even if Windows 10 was perfect software according to the "it gives you the correct output for every input in the input domain" definition, it would still not open the start menu every time you clicked on the icon.

Re: Zig: software should be perfect [video]

#52

Post by electrograv » Sun, Nov 11, 2018, 2:07 AM UTC

bjourne wrote: post #51

electrograv wrote: post #46

Earlier quoted context omitted.

While I agree with the content of your post literally, I think we often underestimate the importance of software reliability and performance, and end up giving it less attention than it deserves. I understand the place for rapid prototyping etc., and that not every software application deals with life-and-death situations — but even those that aren’t, I think our industry suffers a bit here. For example, to this day…

It is entirely possible that even if Windows 10 was perfect software according to the "it gives you the correct output for every input in the input domain" definition, it would still not open the start menu every time you clicked on the icon.

Of course we can redefine bugs as features, but to present that as an argument is a “red herring”: I can assure you, the Windows 10 start menu is not i tended to fail or delay opening instantly upon click or system button press.

Re: Zig: software should be perfect [video]

#53

Post by berti » Sun, Nov 11, 2018, 2:08 AM UTC

clarry wrote: post #22

Tempest1981 wrote: post #7

Definitely nice to have tools (compiler warnings, static analyzers) that can keep you from hurting yourself. Of course many projects don’t use them, for various reasons. I wonder: is turning on full compiler warnings, then fixing them - is it making my software better, or just satisfying some type of neuroticism?

There are stupid warnings about nothing to fix, at least in gcc. It's not making the software better. One warning I recently disabled on $workproject is -Wtrigraphs.

Trigraphs can readily be formed unintentionally, so having a warning when they change the meaning of the program seems valuable. There is a good reason they were removed entirely in C++17.

Re: Zig: software should be perfect [video]

#54

Post by dmpk2k » Sun, Nov 11, 2018, 2:11 AM UTC

bjourne wrote: post #51

electrograv wrote: post #46

Earlier quoted context omitted.

While I agree with the content of your post literally, I think we often underestimate the importance of software reliability and performance, and end up giving it less attention than it deserves. I understand the place for rapid prototyping etc., and that not every software application deals with life-and-death situations — but even those that aren’t, I think our industry suffers a bit here. For example, to this day…

It is entirely possible that even if Windows 10 was perfect software according to the "it gives you the correct output for every input in the input domain" definition, it would still not open the start menu every time you clicked on the icon.

But less likely.

It's clear what his point is.

Re: Zig: software should be perfect [video]

#55

Post by Jach » Sun, Nov 11, 2018, 2:18 AM UTC

loeg wrote: post #49

sqrt17 wrote: post #42

Earlier quoted context omitted.

There are still plenty of exceptions that are unchecked, i.e. subtypes of RuntimeException, which do not have to be declared on the callee. As well as people who think it's a good idea to throw instances of Exception and just tack a "throws Exception" on their methods.

Right — that's a failing of Java the language; not the concept of checked exceptions. At least one problem with Java's exception model that Zig does not have is that some exceptions are "unchecked."

The 'failing' of checked exceptions comes from essentially being forced to couple the type signatures of your methods' successful results with their error results. A more explicit way to do this is with Optional/Either types, and now you don't need the checked exceptions feature nor need to get people to remember to check for a global errno or some other data convention like an empty string / null. There's a lot of boilerplate though, just like with checked exceptions.

I prefer a more decoupled/late-binding approach to error handling; so far Common Lisp does it the best I've seen.[0] The key insight CL people had was that often in the face of an error, you need help from somewhere away from your local context to figure out how to resolve it, but then for many cases, you want to return back to where the error has occurred and take the resolution path with your local context. In other languages that automatically unwind the stack to the exception handler, it's too late.

[0] http://www.nhplace.com/kent/Papers/Condition-Handling-2001.h...

Re: Zig: software should be perfect [video]

#56

Post by int_19h » Sun, Nov 11, 2018, 2:23 AM UTC

platz wrote: post #30

AndyKelley wrote: post #24

Earlier quoted context omitted.

I made this argument in the talk: the only problem with exception-based handling is the lack of explicitness. It's too easy to call functions without being aware of the set of possible errors. Many c++ projects disable exceptions entirely. Writing "exception safe" code is tricky and non-obvious. Functions which should be guaranteed to never fail often can throw std::bad_alloc. Try-catch syntax forces incorrect nestin…

> the only problem with exception-based handling is the lack of explicitness - It's too easy to call functions without being aware of the set of possible errors. i.e. checked-exceptions?

Once you have checked exceptions, you're basically as verbose as the Error monad anyway. Except that instead of properly encoding it in the function result type, you have a separate mechanism, which is kinda sorta but not quite like a type. In Java, this manifests itself in things such as it being impossible to write a higher-order function that has a signature of "takes some function F, and can throw everything that F can throw, plus some E".

Re: Zig: software should be perfect [video]

#57

Post by ajennings » Sun, Nov 11, 2018, 2:31 AM UTC

I've wanted a language like this. Java's checked exceptions with some way to offload the bookkeeping to the compiler.

What about other run-time exceptions, like divide by zero? Are they checked?

What about Hoare's billion-dollar mistake (null pointer exceptions)? Does Zig have non-nullable references?

Re: Zig: software should be perfect [video]

#58

Post by Scarbutt » Sun, Nov 11, 2018, 2:38 AM UTC

electrograv wrote: post #46

mschwaig wrote: post #23

The definition of 'perfect software' used in the talk is 'it gives you the correct output for every input in the input domain'. To that end no funny business like hidden allocations or hidden control flow should happen behind your back, because an out of memory error or some exception your code does not deal with explicitly would not be a correct output according to that definition. Of course you do not need that lev…

While I agree with the content of your post literally, I think we often underestimate the importance of software reliability and performance, and end up giving it less attention than it deserves. I understand the place for rapid prototyping etc., and that not every software application deals with life-and-death situations — but even those that aren’t, I think our industry suffers a bit here. For example, to this day…

Really though, what's up with the windows 10 start menu...

Re: Zig: software should be perfect [video]

#59

Post by euyyn » Sun, Nov 11, 2018, 2:39 AM UTC

electrograv wrote: post #46

mschwaig wrote: post #23

The definition of 'perfect software' used in the talk is 'it gives you the correct output for every input in the input domain'. To that end no funny business like hidden allocations or hidden control flow should happen behind your back, because an out of memory error or some exception your code does not deal with explicitly would not be a correct output according to that definition. Of course you do not need that lev…

While I agree with the content of your post literally, I think we often underestimate the importance of software reliability and performance, and end up giving it less attention than it deserves. I understand the place for rapid prototyping etc., and that not every software application deals with life-and-death situations — but even those that aren’t, I think our industry suffers a bit here. For example, to this day…

In over 6 years working in this industry, in projects with anywhere from dozens of users to millions of users, not a single bug I have encountered was caused by a "hidden allocation" causing the process to go OOM. Not one instance.

One of the two examples he cited in his introduction, the Android one, wasn't even caused by an unhandled OOM error. Android by design will kill unused processes if they're occupying memory that the foreground process needs.

If we want software without bugs, removing hidden allocations in languages is far down in the priority list.

Re: Zig: software should be perfect [video]

#60

Post by Jach » Sun, Nov 11, 2018, 2:40 AM UTC

electrograv wrote: post #48

visualstudio wrote: post #45

I'm watching Zig and Jai closely. We need a better C and C++ isn't it. Good luck!

I honestly think Zig has the potential to be the C/C++ replacement. I haven’t checked out Jai yet but will now that you mention, thanks! This is somewhat subjective of course, but from what I’ve seen, Zig has just the right set of features to modernize systems programming, without making the language too complex or difficult to write (which arguably Rust’s “borrow checker” system does), and (like Rust) gets rid of so…

There are a lot of contenders. My own unsorted list is Nim, Rust, C++20xx, D, Objective C, Pony, Zig, Crystal, Red, and maybe a form of Lisp (why not Common Lisp). Even more unlikely a maybe and only for the C/C++ trenches of embedded systems, some form of Forth. If Jai ever ships I might consider adding it (at least as a contender to the C/C++ trenches of games and game engines), but it's absurd to think it will have any impact when it can't even be used by anyone other than jblow yet. Even if it ships, I would bet its highest anywhere-realistic impact (which is still damn high) would be to become the PHP of game programming. The mythical C/C++ replacement that everyone will choose when they previously would have chosen C or C++, causing C/C++ to die like COBOL? Much less likely.