Re: Latest CVS make compile warnings fix

[J. Grant]

Hi Paul,

[...]
>   jg> If I submit fixes for these would they be reviewed for inclusion ?
>
> I'm happy to look at them, but they will not be easy to fix.

I'm happy to assist if needed.  I hope the 3.81 release can have as few
warnings as possible.

>   jg> main.c:420: warning: missing initializer
>   jg> main.c:420: warning: (near initialization for `switches[30].type')
>
> This one, IMO, could be a bug in GCC.  As far as I can tell the
> initializer is correct.  The only way to avoid the warning seems to be
> to put a full suite of 0 initializations into the last element.  But
> that shouldn't be necessary, if I remember my ISO standard properly.
>
>   jg> main.c: In function `main':
>   jg> main.c:1790: warning: comparison of unsigned expression < 0 is always 
> false
>   jg> main.c:2122: warning: comparison of unsigned expression < 0 is always 
> false
>
>   jg> remake.c: In function `update_file_1':
>   jg> remake.c:432: warning: comparison of unsigned expression < 0 is always 
> false
>   jg> remake.c: In function `notice_finished_file':
>   jg> remake.c:863: warning: comparison of unsigned expression < 0 is always 
> fals
>
>   jg> file.c: In function `file_timestamp_cons':
>   jg> file.c:679: warning: comparison of unsigned expression < 0 is always false
>   jg> file.c:680: warning: comparison between signed and unsigned
>   jg> file.c:680: warning: comparison of unsigned expression < 0 is always false
>   jg> file.c:683: warning: comparison of unsigned expression < 0 is always false
>
> All of these are not going to be easy to fix.  The code is actually
> created that way for a reason, IIRC.

It will be a little difficult to work out what the code is that is
compiled, as there seems to be (traditionally) excessive use of
pre-processor macro expansion to inline code.  Could we update to be
inline functions which we can leave the compiler to decide if it that
suits the build config to inline?

>   jg> job.c: In function `start_job_command':
>   jg> job.c:944: warning: variable `flags' might be clobbered by `longjmp' or 
>   jg> `vfork'
>   jg> job.c:948: warning: variable `argv' might be clobbered by `longjmp' or 
>   jg> `vfork'
>   jg> job.c: In function `construct_command_argv_internal':
>
> Hm.  I think this has already been resolved: are you looking at the
> beta3 code??  I don't get this warning anymore in my version.

I did cvs update -d after your last commit, and also just now.  My code
still gives me that warning.  I am using gcc (GCC) 3.3.5 (Debian 1:3.3.5-8)
on on an i586.

There is another similar one here:

function.c: In function `func_shell':
function.c:1466: warning: variable `error_prefix' might be clobbered by
`longjmp' or `vfork'
function.c:1457: warning: argument `o' might be clobbered by `longjmp'
or `vfork

Kind regards
JG