Re: [9fans] quick question about proc.c:error(char *err)

Re: [9fans] quick question about proc.c:error(char *err)

[Russ Cox]

> void
> error(char *err)
> {
>         spllo();
>         kstrcpy(up->errstr, err, ERRMAX);
>         setlabel(&up->errlab[NERR-1]);
>         ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
>         nexterror();
> }

The setlabel is useful if you get an error stack
underflow or some other mismanagement error.
You can (sometimes) look at the label to reconstruct
the stack at the time that error was called.
It's for debugging only.

> What are the conditions when this label might be get used ?

Hopefully never.

> And what is expected from jumping into error ?

Error shouldn't return.  It should go to the innermost
waserror, returning non-zero.  Calling nexterror does this.

Russ

[9fans] quick question about proc.c:error(char *err)

[Roman V. Shaposhnick]

Can someone, please, give me any hint on why the underlined line of code
is needed in error():

void
error(char *err)
{
        spllo();
        kstrcpy(up->errstr, err, ERRMAX);
        setlabel(&up->errlab[NERR-1]);
        ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
        nexterror();
}

What are the conditions when this label might be get used ? And what is
expected from jumping into error ?

Thanks,
Roman.