Re: [gpsd-dev] gpsd style change?

[Eric S. Raymond]

Gary E. Miller <address@hidden>:
> Yo All!
> 
> A lot of changes like this in recent git commit:
>       5b4af8cc002773058ad744f9c2f9975284b09224
> 
> diff --git a/monitor_ubx.c b/monitor_ubx.c
> index 8b4101e..f73a06b 100644
> --- a/monitor_ubx.c
> +++ b/monitor_ubx.c
> @@ -260,13 +260,13 @@ static void ubx_update(void)
>      if (pps_thread_lastpps(&session, &drift) > 0) {
>         /* NOTE: can not use double here due to precision requirements */
>         struct timespec timedelta;
> -        char buf2[22];
>         TS_SUB( &timedelta, &drift.clock, &drift.real);
>          if ( 86400 < (long)labs(timedelta.tv_sec) ) {
>             /* more than one day off, overflow */
>              /* need a bigger field to show it */
>             (void)mvwprintw(ppswin, 1, 13, "> 1 day");
>          } else {
> +           char buf2[22];
>             (void)timespec_str( &timedelta, buf2, sizeof(buf2) );
>             (void)mvwprintw(ppswin, 1, 13, "%s", buf2);
>          }
> 
> This is actually my preferred style of coding: to keep local variables
> as local as possible.
> 
> But, when I have done this in the past, following commits have hoisted
> the variable declarations to the top of the function.

That's odd.  I always move them as close to usage as possible, both
because I share your preference and because cppcheck throws warnings
if the scope of a variable can be reduced.
 
> Is this now an official style change?  I would prefer it that way.  If
> so can it be made official somewhere so things don't move back and
> forth.

We don't have an official style guide, but I am 100% behind giving
variables scopes as narrow as possible. Go with your instinct on this.
 
> Or, bigger picture, maybe a real style guide?

Be careful what you wish for.  If we had one, some of your coding
quirks might be agin' it.  My eyeballs stumble whenever I see that
"if (0 == foo)" thing you like to do...
-- 
                <a href="http://www.catb.org/~esr/";>Eric S. Raymond</a>