Bug 236 – Class literal expression always says "base classes expected"

I never ran into this before, and I kind of wonder why. void delegate() foo() { return &(new class { void dg() { writefln("delegate!"); } }).dg; } This gives the error "base classes expected" for the class literal. I guess I never ran into it because all the class literals I ever wrote derived from other classes. I tried this in several versions of DMD all the way back to 0.126, and they all give the same error. According to the spec, base classes are optional for class literals. I should be able to write "new class {}" and it should be fine. The semantic equivalent works just fine: void delegate() foo() { class fofof { void dg() { writefln("delegate!"); } } return &(new fofof).dg; }

Fixed DMD 0.163.

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 [email protected] schrieb am 2006-07-02: > http://d.puremagic.com/issues/show_bug.cgi?id=236 > I never ran into this before, and I kind of wonder why. > > void delegate() foo() > { > return &(new class > { > > void dg() > { > writefln("delegate!"); > } > }).dg; > } > > This gives the error "base classes expected" for the class literal. I guess I > never ran into it because all the class literals I ever wrote derived from > other classes. I tried this in several versions of DMD all the way back to > 0.126, and they all give the same error. According to the spec, base classes > are optional for class literals. I should be able to write "new class {}" and > it should be fine. > > The semantic equivalent works just fine: > > void delegate() foo() > { > class fofof > { > void dg() > { > writefln("delegate!"); > } > } > > return &(new fofof).dg; > } Added to DStress as http://dstress.kuehne.cn/run/c/class_26_A.d http://dstress.kuehne.cn/run/c/class_26_B.d Thomas -----BEGIN PGP SIGNATURE----- iD8DBQFE4Z86LK5blCcjpWoRAnVaAJ91paqlf+5cmbO8IhSRRAdOHuMnEQCdE8Lp /MUhZsGnq18Le2uppLjHkdE= =z9aB -----END PGP SIGNATURE-----