Re: Newbie question about compiler warnings.
I demand that Neil Williams may or may not have written...
> On Sun, 14 Jan 2007 12:54:17 -0800 Brandon Barnes
> <winterknight@nerdshack.com> wrote:
>> Are we allowed to disable compiler warnings? What is the preferred method,
>> if the code is fine, and would require a huge overhaul to fix?
> Warnings are only a problem if -Werror is also set. If upstream don't use
> -Werror, the warnings may actually be useful in debugging the package
> and/or FTBFS bugs from porters.
Agreed. In particular, I've found gcc's -Wformat=2 and -Wformat-security to
be of use. (Fixing -Wformat-security warnings does tend to cause other
warnings, but that's good.)
> As the humble opinion of one DD, hiding any output that does not stop the
> build is a BadIdea(TM) in most cases - as well as being unnecessary.
For C, -Wno-format-zero-length is probably a good idea if using -Wformat (at
least, it is for gxine where the warnings which it suppresses are noise).
> (Which is why I don't mind commented dh_ lines in rules but that's a
> different thread.)
I'd remove those which I'm sufficiently sure will never be needed for
$PACKAGE.
[snip]
>> If so, should I disable them in Makefile, or debian/rules?
> No. I would recommend leaving them in - you may be grateful for the
> evidence in times to come. Warnings are issued for a reason - until
> upstream fix the code, I'd always recommend to leave the warning in place.
Certain ones should be left in anyway, IMO.
--
| Darren Salt | linux or ds at | nr. Ashington, | Toon
| RISC OS, Linux | youmustbejoking,demon,co,uk | Northumberland | Army
| Kill all extremists!
All is well that ends well.
Reply to: