Bug 5997 – Static arrays with 0 length accepted by compiler

Status
RESOLVED
Resolution
INVALID
Severity
normal
Priority
P2
Component
dmd
Product
D
Version
D2
Platform
Other
OS
Windows
Creation time
2011-05-13T19:25:00Z
Last change time
2015-06-09T05:14:46Z
Keywords
accepts-invalid
Assigned to
nobody
Creator
andrej.mitrovich

Comments

Comment #0 by andrej.mitrovich — 2011-05-13T19:25:14Z
Is there a use case for 0-length static arrays? If there's not a use-case, then this should probably be a compiler error: int[0] logs; // <- disallow this void main() { logs = [4]; } The assignment statement causes a linker error (if you comment it out you won't get any linker errors): /+ test.obj(test) Offset 002DFH Record Type 009D Error 16: Index Range --- errorlevel 1 +/
Comment #1 by dlang-bugzilla — 2011-05-13T23:39:57Z
(In reply to comment #0) > Is there a use case for 0-length static arrays? I believe they can be used as a value type in an associative array to create a set.
Comment #2 by bearophile_hugs — 2011-05-14T02:57:56Z
One use case are variable-length structs: struct MyArray(T) { size_t len; T data[0]; // access methods here, that use data.offsetof } You create such array with a C malloc or GC malloc.
Comment #3 by andrej.mitrovich — 2011-05-14T13:21:39Z
(In reply to comment #1) > (In reply to comment #0) > > Is there a use case for 0-length static arrays? > > I believe they can be used as a value type in an associative array to create a > set. Ah, you're right. I even forgot about opening this topic which mentioned this trick: http://www.digitalmars.com/d/archives/digitalmars/D/A_case_for_valueless_AA_s_133165.html I'm closing this down.