Обсуждение: GCC warning in back branches

Hi,

guc.c: In function ‘RestoreGUCState’:
guc.c:9455:4: error: ‘varsourceline’ may be used uninitialized in this
function [-Werror=maybe-uninitialized]
 9455 |    set_config_sourcefile(varname, varsourcefile, varsourceline);
      |    ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

I propose the attached.

On Mon, Feb 15, 2021 at 02:15:51PM +1300, Thomas Munro wrote:
> guc.c: In function ‘RestoreGUCState’:
> guc.c:9455:4: error: ‘varsourceline’ may be used uninitialized in this
> function [-Werror=maybe-uninitialized]
>  9455 |    set_config_sourcefile(varname, varsourcefile, varsourceline);
>       |    ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
>
> I propose the attached.

We usually don't bother much about compilation warnings in stable
branches as long as they are not real bugs, and these are the oldest
stable ones.  So why here?  I would have patched the top of the
function if it were me, btw.
--
Michael

Michael Paquier <michael@paquier.xyz> writes:
> On Mon, Feb 15, 2021 at 02:15:51PM +1300, Thomas Munro wrote:
>> I propose the attached.

> We usually don't bother much about compilation warnings in stable
> branches as long as they are not real bugs, and these are the oldest
> stable ones.  So why here?  I would have patched the top of the
> function if it were me, btw.

If somebody were running a buildfarm member with recent gcc
and -Werror, we'd pretty much have to fix it.

I'd say the real policy is that we don't worry about
uninitialized-variable warnings from old compiler versions,
on the theory that they're probably compiler shortcomings.
But I'd be inclined to fix anything from a current gcc version.

            regards, tom lane

On Mon, Feb 15, 2021 at 2:35 PM Michael Paquier <michael@paquier.xyz> wrote:
> ... I would have patched the top of the
> function if it were me, btw.

I just copied the way it is coded in master (due to commit fbb2e9a0
which fixed this warning in 11+).