Even if it makes no difference in join.c, silencing the compiler warning seems like a good thing. Everything else in sbase compiles with no warnings.
________________________________________ From: [email protected] <[email protected]> on behalf of Hiltjo Posthuma <[email protected]> Sent: 02 February 2017 21:39 To: hackers mail list Subject: Re: [hackers] [sbase][PATCH] Fixed a couple of uninitialised variable warnings from Clang Probably a false positive, but for ed.c it looks ok to me. The join.c patch does not look ok to me (clang doesn't detect that eprintf() exists the program), so it's a false positive. On Thu, Feb 2, 2017 at 3:43 PM, Simon Cooksey <[email protected]> wrote: > - ed.c > Initialise hline in `static int makeline(char, int)' > > - join.c > Intitalise fileno and fldno in `static struct spec * makespec(char)' > --- > ed.c | 2 +- > join.c | 4 ++-- > 2 files changed, 3 insertions(+), 3 deletions(-) > > diff --git a/ed.c b/ed.c > index 4b28848..ae66b8d 100644 > --- a/ed.c > +++ b/ed.c > @@ -155,7 +155,7 @@ back(int c) > static int > makeline(char *s, int *off) > { > - struct hline *lp; > + struct hline *lp = 0; > size_t len; > char c, *begin = s; > > diff --git a/join.c b/join.c > index d3e2343..6145654 100644 > --- a/join.c > +++ b/join.c > @@ -314,8 +314,8 @@ static struct spec * > makespec(char *s) > { > struct spec *sp; > - int fileno; > - size_t fldno; > + int fileno = 0; > + size_t fldno = 0; > > if (!strcmp(s, "0")) { /* join field must be 0 and nothing else */ > fileno = 0; > -- > 2.11.0 > >
