Re: CVS commit: pkgsrc/devel/cmph

[Aleksej Saushev <asau%inbox.ru@localhost>]

John Marino <netbsd%marino.st@localhost> writes:

> On 10/12/2012 17:35, Alistair Crooks wrote:
>> On Fri, Oct 12, 2012 at 08:01:00AM +0000, John Marino wrote:
>>> Module Name:        pkgsrc
>>> Committed By:       marino
>>> Date:               Fri Oct 12 08:01:00 UTC 2012
>>>
>>> Modified Files:
>>>     pkgsrc/devel/cmph: Makefile
>>>
>>> Log Message:
>>> devel/cmph: Fix build for gcc 4.6+
>>>
>>> Remove -Werror (catches -Wunused-but-set-variable) to fix build.
>>
>> Please stop doing this (removing -Werror) now - it is not fixing
>> anything, and will make things harder in future to spot problems.
>>
>> More acceptable ways to fix these kind of things:
>>
>> 1. check if there's a new version out which might have fixes for this
>>
>> 2. fix the problem, not the reporter of the problem
>>
>> 3. mark the package as broken on your platform
>>
>> 4. if you can't do any of these, then let someone else deal with it
>>
>> Thanks,
>> Alistair
>
>
> David Holland _JUST_ got finished saying this approach is fine.

No, he argued completely differently.

I agree with Alistair. If you do mean that GCC 4.7 can't handle it,
implement your hack as a hack, for that particular version of compiler,
and let others fix it properly rather than sweeping problem under the rug
_for_everyone_else_.

> YOUR 1.  That's a huge effort for a non-problem.
> YOUR 2.  Again, it's not a problem.  It's a developer warning.
> It's like having "ok = function(x);" and then never referencing
> "ok" again. It doesn't hurt anything.  The compiler optimizes it
> out.
> YOUR 3.  Absurd.  Also, it's not a DragonFly problem.
> YOUR 4.  Again, absurd.  Nobody is going to touch these for
> years because it doesn't affect them.  pkgsrc is very selfish
> that way, so this suggestion is helpful to nobody.
>
> It is __NOT__ making it harder to spot "problems".  The warning
> is still emitted.  You still see it in logs.  It just makes it
> so the build doesn't stop.
>
> If you continue to insist that DragonFly and all other platforms
> using gcc 4.6+ should remain broken for a no-good reason, I
> don't know what to do.  There are global "hacks" that could be
> set, but it's hardly worse.
>
> Also, those are all of the -Werror type of problems barring the
> last 800 packages I'm running now.  I'm finished fixing this
> type.

There may be lots of other significant warnings that do point to problems,
and those warnings are exactly the reason why -Werror exists and turned on.


-- 
HE CE3OH...