The new GOTO Statement?
-
Yes, you can abuse lambda's and make a mess. You can abuse goto and make a mess. You can abuse boolean variables to simulate some forms of goto and make just as big a mess. You can abuse operator overloading and make a mess, and when James Gosling says you can't have operator overloading in Java, you can make just as big a mess with method overloading and virtual methods if you set your mind to it. You can abuse switch in atrocious ways in C and C++ (case goes pretty much anywhere, it doesn't even look like valid syntax but it is), which is like a goto where you don't even know for sure where it will go. You can abuse pretty much every aspect of a general purpose programming language.
-
After coming down off of the "That's pretty cool" factor and as members on my team have increasingly been using anonymous methods, lambda expressions and new Func<> routines embedded in methods. The complexity (IMO) as increased significantly. I've begun to question this seemingly popular approach as; What's the difference between using embedded functions and a goto statement? It seems to me it's no different and just as difficult to follow and maintain. I'd be curious on other opinions of this practice.
-
Yes, I agree. Every element of a language has it's use - even
goto
andvar
in C#- it's just that if you use them inappropriately you get less readable code instead of more. Personally, I find lambdas are useful in their place, but I avoid using them most of the time.var
should be banned outside Linq!Ideological Purity is no substitute for being able to stick your thumb down a pipe to stop the water
-
After coming down off of the "That's pretty cool" factor and as members on my team have increasingly been using anonymous methods, lambda expressions and new Func<> routines embedded in methods. The complexity (IMO) as increased significantly. I've begun to question this seemingly popular approach as; What's the difference between using embedded functions and a goto statement? It seems to me it's no different and just as difficult to follow and maintain. I'd be curious on other opinions of this practice.
I don't know if the complexity increases significantly, but people well versed in imperative languages, like the majority of C#, may not be familiar with functional programming constructs, making it a little tougher to figure out just from lack of exposure. And they are harder to debug, can't seem to put breakpoints inside of them. But I probably abuse them a little, some sections of my code look more like LISP than a C-family language...especially when I realized I could use them to make wrappers out of certain loops and similar constructs that were being used over and over again in my code...
-
After coming down off of the "That's pretty cool" factor and as members on my team have increasingly been using anonymous methods, lambda expressions and new Func<> routines embedded in methods. The complexity (IMO) as increased significantly. I've begun to question this seemingly popular approach as; What's the difference between using embedded functions and a goto statement? It seems to me it's no different and just as difficult to follow and maintain. I'd be curious on other opinions of this practice.
sisnaz wrote:
What's the difference between using embedded functions and a goto statement?
Because one is cool and the other isn't. It is of course related to the difference between understanding the syntax of a language and being able to write syntax that is maintainable.
-
Yes, you can abuse lambda's and make a mess. You can abuse goto and make a mess. You can abuse boolean variables to simulate some forms of goto and make just as big a mess. You can abuse operator overloading and make a mess, and when James Gosling says you can't have operator overloading in Java, you can make just as big a mess with method overloading and virtual methods if you set your mind to it. You can abuse switch in atrocious ways in C and C++ (case goes pretty much anywhere, it doesn't even look like valid syntax but it is), which is like a goto where you don't even know for sure where it will go. You can abuse pretty much every aspect of a general purpose programming language.
harold aptroot wrote:
You can abuse switch in atrocious ways in C and C++ (case goes pretty much anywhere, it doesn't even look like valid syntax but it is)
Are you saying that you can have a case statement without an enclosing switch? :wtf: What does THAT look like, and what would one use it for?
-
Yes, I agree. Every element of a language has it's use - even
goto
andvar
in C#- it's just that if you use them inappropriately you get less readable code instead of more. Personally, I find lambdas are useful in their place, but I avoid using them most of the time.var
should be banned outside Linq!Ideological Purity is no substitute for being able to stick your thumb down a pipe to stop the water
-
sisnaz wrote:
What's the difference between using embedded functions and a goto statement? It seems to me it's no different and just as difficult to follow and maintain. I'd be curious on other opinions of this practice.
:doh: A GoTo breaks the flow where as a embedded function becomes part of the flow. Here is some psuedo code to show you the difference
bool flag = GetFlag() //Embedded but the point is made elsewhere
if flag
GOTO: SomeLabelEmbeddedMethod();
SomeOtherEmbeddedMethod();
OK, so this code will ONLY run the
EmbeddedMethod
if theflag
is false. One would think in that case it will also run theSomeOtherEmbeddedMethod
and here is lies the evil of GoTo. What if theEmbeddedMehtod
is defined in the same manner using a GoTo? i.e.bool flag = GetSomeOtherFlag()
if flag
GoTo: SomeOtherLabelNow as a user when I am looking at the first code snippet I have to account for that extra GoTo that could happen. Maybe I must run some logic if the
EmbeddedMethod
runs. How do I ensure it in the most simple manner? Solution, Dont't use GoTo!Computers have been intelligent for a long time now. It just so happens that the program writers are about as effective as a room full of monkeys trying to crank out a copy of Hamlet.
Perhaps embedded method is a loose term. This is what I'm referring to. In my opinion it reflects the same goto example you posted.
public string ReturnSomething()
{
// ... some logic
// ...
var compare = new Func<string, string, string, string, bool>((compare1, compare2, compare3, compare4) =>
{
return (compare1.Equals(compare2, StringComparison.InvariantCultureIgnoreCase) &&
compare3.Equals(compare4, StringComparison.InvariantCultureIgnoreCase));
});// some more logic flow // .... if (compare("a", "b", "c", "d")) { // some logic } return "Something"; }
-
Yes, I agree. Every element of a language has it's use - even
goto
andvar
in C#- it's just that if you use them inappropriately you get less readable code instead of more. Personally, I find lambdas are useful in their place, but I avoid using them most of the time.var
should be banned outside Linq!Ideological Purity is no substitute for being able to stick your thumb down a pipe to stop the water
OriginalGriff wrote:
var
should be banned outside Linq!Oh you CAN use them elsewhere!
Never underestimate the power of human stupidity RAH
-
Actually I liked GOSUB but now I know why!
Never underestimate the power of human stupidity RAH
-
Yes, I agree. Every element of a language has it's use - even
goto
andvar
in C#- it's just that if you use them inappropriately you get less readable code instead of more. Personally, I find lambdas are useful in their place, but I avoid using them most of the time.var
should be banned outside Linq!Ideological Purity is no substitute for being able to stick your thumb down a pipe to stop the water
I, for one, welcome our new
var
overlord. Use it all the time, except in non-assignment declarations. -
Perhaps embedded method is a loose term. This is what I'm referring to. In my opinion it reflects the same goto example you posted.
public string ReturnSomething()
{
// ... some logic
// ...
var compare = new Func<string, string, string, string, bool>((compare1, compare2, compare3, compare4) =>
{
return (compare1.Equals(compare2, StringComparison.InvariantCultureIgnoreCase) &&
compare3.Equals(compare4, StringComparison.InvariantCultureIgnoreCase));
});// some more logic flow // .... if (compare("a", "b", "c", "d")) { // some logic } return "Something"; }
I fail to see anything like goto in your example. Flow branching into a function when it is called is all I see, and that happens all over, every second in C#. The key difference here is that without using goto in your function body, you are still guaranteed a return to just after the line that calls the function.
-
Yes, you can abuse lambda's and make a mess. You can abuse goto and make a mess. You can abuse boolean variables to simulate some forms of goto and make just as big a mess. You can abuse operator overloading and make a mess, and when James Gosling says you can't have operator overloading in Java, you can make just as big a mess with method overloading and virtual methods if you set your mind to it. You can abuse switch in atrocious ways in C and C++ (case goes pretty much anywhere, it doesn't even look like valid syntax but it is), which is like a goto where you don't even know for sure where it will go. You can abuse pretty much every aspect of a general purpose programming language.
Then some guru comes along and declares any one of those language features dangerous and unsafe. The followers of the guru create a cult around that and go a little further by creating the sacred commandment 'Thou shalt not use (whatever)'. The following crusade will take years, but in the end we have nagging code checking tools that assume to know better than that overpaid code monkey in front of the machine.
harold aptroot wrote:
You can abuse pretty much every aspect of a general purpose programming language.
That's exactly why the gurus will never run out of work. In the wrong direction, if you ask me. Coders who have been trained to follow rules blindly are very helpless when something goes wrong. That costs time and money instead of making anything safer or more productive. If you, my special Java friends, happen to read this: Your current problems will end the day you finally start thinking, throw out the holy commandment not to manage memory and analyze your code. Or you can spend some more years trying to beat the garbage collection into submission. :)
At least artificial intelligence already is superior to natural stupidity
-
Bad Software languages breed bad software engineers....
Computers have been intelligent for a long time now. It just so happens that the program writers are about as effective as a room full of monkeys trying to crank out a copy of Hamlet.
-
var
is great for foreach loops when you have long types like KeyValuePairs with a generic type for the value. I don't use it often outside that.It's easier to type - but it doesn't make the code more readable! You have to check the type of the IEnumerable that the variable is declared from in order to find out what type you are using within the loop:
foreach (var v in MyClass.Items)
{
...
}Or
foreach (KeyValuePair<Guid, List<string>> kvp in MyClass.Items)
{
...
{(Not that I'm advocating using
KeyValuePair<Guid, List<string>>
directly anyway, you understand)Ideological Purity is no substitute for being able to stick your thumb down a pipe to stop the water
-
Yes, I agree. Every element of a language has it's use - even
goto
andvar
in C#- it's just that if you use them inappropriately you get less readable code instead of more. Personally, I find lambdas are useful in their place, but I avoid using them most of the time.var
should be banned outside Linq!Ideological Purity is no substitute for being able to stick your thumb down a pipe to stop the water
I quite like the use of 'var' when the thing to the right is clearly typed, e.g.
var map = new Dictionary<string, int>();
It was rather a failing in all C family languages that you had to write the type twice before when it's clearly there in the initialiser, and using var here is not hiding anything. I don't actually like it in Linq, it's too hard to see by inspection what the type of a Linq expression is. I quite like declaring those as IQueryable<whatever>.
-
harold aptroot wrote:
You can abuse switch in atrocious ways in C and C++ (case goes pretty much anywhere, it doesn't even look like valid syntax but it is)
Are you saying that you can have a case statement without an enclosing switch? :wtf: What does THAT look like, and what would one use it for?
Well it doesn't really go that far, but the case statements don't define any sort of block (they behave much the same way as labels) and (as long as it's somewhere in a switch) you can mix them with other control flow. For example, Duff's device[^], which mixes a switch and a do/while.
-
Perhaps embedded method is a loose term. This is what I'm referring to. In my opinion it reflects the same goto example you posted.
public string ReturnSomething()
{
// ... some logic
// ...
var compare = new Func<string, string, string, string, bool>((compare1, compare2, compare3, compare4) =>
{
return (compare1.Equals(compare2, StringComparison.InvariantCultureIgnoreCase) &&
compare3.Equals(compare4, StringComparison.InvariantCultureIgnoreCase));
});// some more logic flow // .... if (compare("a", "b", "c", "d")) { // some logic } return "Something"; }
compare is nothing more or less than a local method; this example is no less readable than the boring alternative:
public string ReturnSomething()
{
// ... some logic
// ...// some more logic flow // .... if (compare("a", "b", "c", "d")) { // some logic } return "Something"; }
private boolean compare(string compare1, string compare2, string compare3, string compare4) {
return (compare1.Equals(compare2, StringComparison.InvariantCultureIgnoreCase) &&
compare3.Equals(compare4, StringComparison.InvariantCultureIgnoreCase));
}... and arguably more so, if the method is only used in one place, because you're not gumming up the class scope with methods that are not relevant to any of it apart from one function. compare cannot affect the control flow of the containing function, it is a normal function which takes arguments and returns a result – it is just declared as a dynamic Func type for technical reasons. Delphi allows you to declare local methods statically for exactly this kind of situation, and (like so much in Delphi) that is a really good idea.
-
YES! I completely agree. I have a team member that declares EVERYTHING as var. He says it's because it makes it loosely coupled and also C Sharpner tells him too.
haha sounds like me. var is certainly not only nice for foreach loops or LINQ. I find it rather stupid to type something like:
Dictionary>> stuff = /*sigh*/ new Dictionary>>();
Yes, stupid indeed. And it's exactly that type of scenario the "var" keyword was made for.
-
Well it doesn't really go that far, but the case statements don't define any sort of block (they behave much the same way as labels) and (as long as it's somewhere in a switch) you can mix them with other control flow. For example, Duff's device[^], which mixes a switch and a do/while.