Anthony Romano

I know, but I want to link with an existing network that's all inspircd. I'm assuming it'll buy some amount of TS6 scaffolding that can be reused to support more...

@gyuho it's important for current delete semantics: ```go // gofail: var crash struct{} // panic("crash") ``` if the lock isn't held for the duration of the fp, then it's possible...

Failpoints by convention only set a single value-- they're not functions. What's your use case?

Marshall it into a string: ``` func whatever() (*someStruct, error) { ... // gofail: var someData string // return unmarshalMyData(someData) ... } func unmarshalMyData(v string) (*someStruct, error) { ... }...

@gap892003 can you describe what this patch is supposed to fix? The bindings should already be removed on `disable` when the `.fail.go` files are removed (see https://github.com/coreos/gofail/blob/master/gofail.go#L190)

@gap892003 could you please provide some example code to show what's going on? It's not clear to me why the imports are sticking around.

@gap892003 OK, that seems unreliable if it needs an IDE to restore the import when re-enabling. Doing this kind of pattern with build tags instead might be cleaner: something.go: ```go...

@gap892003 it would need a separate `// +build failpoint` file for each package using unimported failpoints. Usually a failpoint shouldn't have much code in the first place (e.g., return an...

@giudon this doesn't look like a bug; the code allocates 1GB chunks after the first 1GB. If given a 1GB minimum size, it should return 1GB. If given a 1.01GB...

@pinfake is there a simple program that can reproduce this issue?