Reduce code complexity: Guard clauses

[Greg Utas]

It's brief and recommends putting guards in their own if statements. It will trigger those who cleave to the shibboleth that a function should only have one return statement.

Robust Services Core | Software Techniques for Lemmings | Articles
The fox knows many things, but the hedgehog knows one big thing.

[realJSOP]

I cleave to that shibboleth.

".45 ACP - because shooting twice is just silly" - JSOP, 2010
-----
You can never have too much ammo - unless you're swimming, or on fire. - JSOP, 2010
-----
When you pry the gun from my cold dead hands, be careful - the barrel will be very hot. - JSOP, 2013

[den2k88]

Like PC-Lint and MISRA standards.

GCS d--(d-) s-/++ a C++++ U+++ P- L+@ E-- W++ N+ o+ K- w+++ O? M-- V? PS+ PE- Y+ PGP t+ 5? X R+++ tv-- b+(+++) DI+++ D++ G e++ h--- r+++ y+++*      Weapons extension: ma- k++ F+2 X

[Richard Deeming]

Quis custodes custodit? :-D

---

"These people looked deep within my soul and assigned me a number based on the order in which I joined." - Homer

[Gary Wheeler]

A completely stupid idea. It adds a thrown exception that functions solely as a goto.

Software Zen: delete this;

[Gary Wheeler]

Agreed.

Software Zen: delete this;

[Marc Clifton]

It's so sad that other programmers that I may end up working with at some point read that drivel and adopt it.

Latest Articles:
DivWindow: Size, drag, minimize, and maximize floating windows with layout persistence

[Kent Sharkey]

At least with a GOTO you can get back to where you were - much more difficult with exceptions. :sigh:

TTFN - Kent

[Gary Wheeler]

True, but then I'm a purist. I haven't used a goto statement in over 20 years. I will admit I use break and continue, but they are 'structured' to my way of thinking.

Software Zen: delete this;

[Greg Utas]

Did I miss something? An exception is more than a goto, because it can jump outside of the current scope. This is very useful for aborting work that has run into a serious error, when it is appropriate to exit a chain of function calls and recover from a known point well down the stack. That's actually the only thing I use exceptions for. In most cases, such as a bad argument, the function should simply return a failure result instead. EDIT: It looks like the code could well be overusing exceptions when returning a failure result would be more appropriate, so that's probably what I missed.

Robust Services Core | Software Techniques for Lemmings | Articles
The fox knows many things, but the hedgehog knows one big thing.

[Gary Wheeler]

The 'guard' code from the article was this:

fun guard() {
if (!condition1) {
throw Exception1()
}
doThings1()

if (!condition2) {
    throw Exception2()
}
doThings2()

if (!condition3) {
    throw Exception3()
}
doThings3()

}

This call needs to be surrounded by a try...catch in order to exactly reproduce the functionality of the original. I'm not sure which language is used for the original code; I think it may be Javascript, but the fun keyword is odd. In any case, throwing an exception in lieu of nested conditionals or a goto is overly complicated and error-inducing. Imagine debugging this sort of nonsense when the catch misses one of the exceptions thrown.

Software Zen: delete this;

[Daniel Pfeffer]

:omg:

Freedom is the freedom to say that two plus two make four. If that is granted, all else follows. -- 6079 Smith W.

[BillWoodruff]

Hi Marc, Perhaps my memory is flawed, but, I seem to recall eyeballing code of yours that used preventive filters in this way. I'd appreciate your comments on my post below !

«The mind is not a vessel to be filled but a fire to be kindled» Plutarch

[BillWoodruff]

I believe in anticipating/handling errors in parameter data before you get to the juicy parts, and I think guard clauses are a good example of a strategy for that. I also think liberal use of Try/Catch for things like file-ops is a good thing. Guard clauses' "virtual isolation" at the start of a method is, imho, "separation of concerns." You can easily comment them out, remove them, as conditions/state management require. However, there are other strategies. What I'd like to see is a radical extension of Attributes in C# that do AOP-like transformations, and, internally generate the equivalent of guard clauses: there's already an example of that in the semantics of the Nullable attributes: [^]. A "wedding" of unit-tests and attributes :)

«The mind is not a vessel to be filled but a fire to be kindled» Plutarch

[Richard Andrew x64]

BillWoodruff wrote:

What I'd like to see is a radical extension of Attributes in C#

You always did strike me as a radical!

The difficult we do right away... ...the impossible takes slightly longer.