Timeline for Readability a=b=c or a=c; b=c;?
Current License: CC BY-SA 2.5
9 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Feb 2, 2014 at 19:51 | comment | added | Ben Voigt |
@Mr.Smith: That is an interesting (and correct) observation, but c is not a property so it doesn't apply here.
|
|
| Feb 2, 2014 at 19:50 | comment | added | Mr. Smith |
strictly speaking, a=b=c=d; in C# is not the same as writing c=d;b=c;a=b;; this is apparent when c is a property (its get won't be invoked with a=b=c=d; but it will with c=d;b=c;a=b).
|
|
| Mar 21, 2011 at 14:33 | comment | added | Ben Voigt |
@Martin: POD-ness has not much to do with it. A type can be POD and still have a user-defined operator=. Anyway, if the operator= is user-defined and not inline, the compiler can't reorder, but in the common case reordering is possible.
|
|
| Mar 21, 2011 at 6:59 | comment | added | Loki Astari | @Ben Voigt : With the assumption that the variables are POD. If not then the compiler can re-order the assignments. So unless you want to have different style one for POD and another for class types you leave yourself open to the confusion of the syntax. | |
| Mar 21, 2011 at 4:56 | comment | added | Ben Voigt | @Martin: Again, unless some of the variables are volatile, the compile can reorder the assignments. So which order is really "equivalent" is more a matter of pedanticism than anything. | |
| Mar 21, 2011 at 4:11 | comment | added | Loki Astari | The fact that you have to point out that the assignment order is not obvious, should be a recommendation not to use it. | |
| Mar 21, 2011 at 2:37 | comment | added | Ben Voigt |
@Paulo: Good point. I'm specifically listing the behavior of C++, but I thought the other languages retain compatibility in this area. You're right that b=c vs b=0 only matters if some of the variables are volatile, the main point I was making was that c gets assigned before a, although again that should make a difference except for volatile, since the compiler can do reordering of writes to normal variables.
|
|
| Mar 21, 2011 at 2:24 | comment | added | Paŭlo Ebermann |
If you do such assertions on a multiple-language question, please specify for which language they are valid. I think in Java it would be c = 0; b = 0; a = 0; instead. (The differences would only be observable if using volatile variables, though.)
|
|
| Mar 21, 2011 at 2:14 | history | answered | Ben Voigt | CC BY-SA 2.5 |