[Libreoffice-qa] minutes of ESC call ...

[Stephan Bergmann]

On 09/06/2012 06:55 PM, Michael Stahl wrote:
> On 06/09/12 17:42, Michael Meeks wrote:
> i don't have an opinion whether --enable-dbgutil should by default
> enable symbols or not.

First of all, note that --enable-symbols is a misnomer.  It is not 
per-se about any kind of symbols (like not stripping internal symbols on 
ELF, etc.), but rather about debugging information.  (With GCC it 
enables -g "Produce debugging information" 
[<http://gcc.gnu.org/onlinedocs/gcc-4.7.1/gcc/Debugging-Options.html#Debugging-Options>] 
(or -g2, -ggdb2), with MSVC it enables /Zi "enable debugging 
information" ["cl /?" output].)

My main concern for assert-enabled (i.e., --enable-dbgutil) builds is 
that if they abort, it is easy to get a decent backtrace.  For that, it 
is important that internal symbols are not stripped, but much less so 
that --enable-symbols--controlled debugging information is available.

Therefore, I have no decided opinion either on whether --enable-dbgutil 
should imply --enable-symbols by default.

Stephan