-Og in dbgutil builds

[Michael Stahl]

On 06.04.2017 10:42, Miklos Vajna wrote:
> Hi,
> 
> On Thu, Apr 06, 2017 at 09:09:54AM +0100, Caolán McNamara <caolanm at redhat.com> wrote:
>> editeng/source/editeng/impedit2.cxx:1820
>>
>> (gdb) print rType
>> $1 = <optimized out>
>> (gdb) print rTypes
>> $2 = <optimized
>> out>
>> (gdb) print pParaPortion->aScriptInfos
>> value has been optimized out
>> (
>> gdb) print pParaPortion
>> $3 = <optimized out>
>>
>> this Og thing can be quite the pain
> 
> Yesterday I brought up an other problem on IRC which is also specific to
> -Og:
> 
> 14:15 <@vmiklos> mst_: today's HAVE_GCC_OG annoyance: in vcl/source/gdi/pdfwriter_impl.cxx:10871, i put an if (nObject == 15)\nSAL_DEBUG("foo"); and it never appears. tweak config_host.mk 
>                  to change _OG=TRUE to _OG=, rebuild the source file -> it works. it would be really nice to have that -Og thing as an opt-in feature, not on by default
> 14:16 < tml__> vmiklos: is there any good reason for the compiler to assume that nObject can never be 15?
> 14:16 <@sberg> vmiklos, that looks more like a bug somewhere than like genuine -Og fallout?
> 14:17 <@vmiklos> the above code location is a SAL_INFO() line, if i set SAL_LOG accordingly, then i see that nObject *is* 15 once
> 14:18 < tml__> weird
> 
> Michael S: could you please consider making -Og an opt-in feature, not a
> default one?

op-in feature probably means just reverting it, you can always use
CXXFLAGS to set it...

but i'm mostly interested in Norbert's opinion, since the biggest
benefit of -Og is faster tinderbox/jenkins builds.

revert it? or add some --enable-Og-for-the-jenkinses flag?