Wine logging affects `zig build test` when `enable_wine = true`

[playday3008]

Zig Version

0.14.0

Steps to Reproduce and Observed Behavior

• zig init
• zig build test -Dtarget=native-windows -fwine

Test will not succeed as wine uses stderr for logging

Expected Behavior

Test should succeed.

Proposed solution

Set WINEDEBUG as -all while running tests, which disables all wine logging and allows test to succeed.

For example, command below will pass test just fine:

• WINEDEBUG=-all zig build test -Dtarget=native-windows -fwine

Possible solution

Modify this code:

zig/lib/std/Build/Step/Run.zig

Lines 1060 to 1067 in 4d79806

	.wine => |bin_name| {
	if (b.enable_wine) {
	try interp_argv.append(bin_name);
	try interp_argv.appendSlice(argv);
	} else {
	return failForeign(run, "-fwine", argv[0], exe);
	}
	},

by adding

        if (run.stdio == .zig_test) {
            try b.graph.env_map.put("WINEDEBUG", "-all");
        }

right after

.wine => |bin_name| {
    if (b.enable_wine) {

and it works, I tested, but I'm not sure whether it's a good solution.

Full snippet

.wine => |bin_name| {
    if (b.enable_wine) {
        if (run.stdio == .zig_test) {
            try b.graph.env_map.put("WINEDEBUG", "-all");
        }
        try interp_argv.append(bin_name);
        try interp_argv.appendSlice(argv);
    } else {
        return failForeign(run, "-fwine", argv[0], exe);
    }
},

[alexrp]

It does succeed; check the exit code. Output on standard error does not by itself cause a build failure. That's why it's marked as stderr in the build summary, as opposed to failure.

I think there's potentially an argument to be made for us setting WINEDEBUG=-all, but only if that environment variable isn't already present and set to some other value.

[playday3008]

It does succeed; check the exit code. Output on standard error does not by itself cause a build failure. That's why it's marked as stderr in the build summary, as opposed to failure.

True, I didn't notice it earlier, just ran it through VS Code and test succeeds, it just kind of misleading, because you know, usually red text == bad.

I think there's potentially an argument to be made for us setting WINEDEBUG=-all, but only if that environment variable isn't already present and set to some other value.

How about:

if (run.stdio == .zig_test) {
    if (b.graph.env_map.get("WINEDEBUG")) |_| {} else {
        try b.graph.env_map.put("WINEDEBUG", "-all");
    }
}

(still don't know if that's the best solution, I'm not that familiar with the whole build system)