Changeset 1372

[PaulBlay]

This line causes a compiler error (well, three errors to be precise).

u32b dupcheck[z_info->e_max];

error C2057: expected constant expression
error C2466: cannot allocate an array of constant size 0
error C2133: 'dupcheck' : unknown size

[EDITx2] It compiles with Dev-C++ (but not with VC++?). Going to check some more things ...

[EDITx3] Pretty sure it should be the following code (but I have an unrelated problem that is Angband/65 only. :-(

	u32b *dupcheck;
	dupcheck = C_ZNEW(z_info->e_max, u32b);

[Pete Mack]

That array initialization is a C99-ism. Your fix will solve it, if you really want to. Don't forget to free dupcheck when you are done with it.

[Magnate]

That array initialization is a C99-ism. Your fix will solve it, if you really want to. Don't forget to free dupcheck when you are done with it.

This causes me a problem. I wrote that code, and it compiled cleanly on gcc with no warnings whatsoever. But I know that Angband is supposed to be C89-compliant, because ajps and others often correct my code to C89 standards. Is there a way that I can tell gcc to be strict about C89?

@paul: thanks for the tip - I'll correct it tonight. I didn't think you had to use C_ZNEW if you were throwing the array away at the end of the function (which we are in this case).

[Magnate]

This causes me a problem. I wrote that code, and it compiled cleanly on gcc with no warnings whatsoever. But I know that Angband is supposed to be C89-compliant, because ajps and others often correct my code to C89 standards. Is there a way that I can tell gcc to be strict about C89?

Come to think of it, since I invoke gcc via make, shouldn't whatever does this go into the Angband makefiles?

[takkaria]

Come to think of it, since I invoke gcc via make, shouldn't whatever does this go into the Angband makefiles?

I think the problem here is that the autoconf buildsystem don't have the same flags as Makefile.std does, the latter of which I use for most development. I'll have a look into transferring the flags across.

[Magnate]

Fixed in r1374. Thanks Paul.

[Nick]

That array initialization is a C99-ism. Your fix will solve it, if you really want to. Don't forget to free dupcheck when you are done with it.

Ah, that explains why stuff like that didn't compile cleanly for me a while ago (probably for WinCE in eVC++). Keep it up, Pete - I'll be a programmer one day.

[Pete Mack]

This causes me a problem. I wrote that code, and it compiled cleanly on gcc with no warnings whatsoever. But I know that Angband is supposed to be C89-compliant, because ajps and others often correct my code to C89 standards. Is there a way that I can tell gcc to be strict about C89?

From http://gcc.gnu.org/onlinedocs/gcc-3....Standards.html
Possibilities include:

-std=c89 or -std=iso9899:1990

Don't apply this uniformly on the OSX build, because main-crb.c requires std=g99, as well as objective C.