[Issue 18977] New: struct and class declarations are inconsistent about what happens when they're marked with immutable

d-bugmail at puremagic.com d-bugmail at puremagic.com
Wed Jun 13 20:45:56 UTC 2018


https://issues.dlang.org/show_bug.cgi?id=18977

          Issue ID: 18977
           Summary: struct and class declarations are inconsistent about
                    what happens when they're marked with immutable
           Product: D
           Version: D2
          Hardware: All
                OS: All
            Status: NEW
          Severity: normal
          Priority: P1
         Component: dmd
          Assignee: nobody at puremagic.com
          Reporter: issues.dlang at jmdavisProg.com

This code

----
immutable struct S {}
immutable class C {}

void main()
{
    S s;
    pragma(msg, typeof(s));

    C c;
    pragma(msg, typeof(c));
}
----

prints

immutable(S)
C

I find this very surprising. It is my understanding that marking a struct or
class as immutable only affects its members, not the type itself. The type
doesn't magically become immutable everywhere it's used - and having it be
magically immutable everywhere is going to be really confusing. I don't see how
anyone can expect to look at

S s;

and think anything other than s is mutable, so it strikes me as a terrible idea
to have it be immutable based on the fact that the struct declaration was
marked with immutable

So, I would strongly argue that this should be fixed so that s is not
implicitly immutable and acts more like what happens with the class case.
However, regardless of whether the struct variable should be implicitly
considered immutable or not, the fact that it is while the class variable is
not is inconsistent, which makes what's going on harder to understand and more
confusing.

So, either, we should fix it so that marking a struct declaration as immutable
has no effect on the mutability of any variables of that type (just its
members), or we should fix it so that marking a class declaration as immutable
makes all variables of that type immutable just like it currently does with
structs. I vote for making it so that variables are never implicitly immutable
based on the declaration of the types, but either way, it should be consistent.

--


More information about the Digitalmars-d-bugs mailing list