<html>
<head>
<meta http-equiv="Content-Type" content="text/html; charset=UTF-8">
</head>
<body>
<div class="moz-cite-prefix">Le 28/09/2021 à 16:04, Jacek Caban a
écrit :<br>
</div>
<blockquote type="cite"
cite="mid:6c69bbe1-1044-c9bb-c949-16e5eae8853e@codeweavers.com">Hi
Eric,
<br>
<br>
On 9/28/21 3:44 PM, Eric Pouech wrote:
<br>
<blockquote type="cite">@@ -2040,9 +2044,10 @@ char*f(const char
*h,char n) {return strchr(h,n);}]])],[ac_cv_c_logicalop_noisy=
<br>
esac
<br>
dnl Default to dwarf-2 debug info
<br>
+ AC_SUBST(DEBUGFORMAT)
<br>
for ac_flag in $CFLAGS; do
<br>
case $ac_flag in
<br>
- -g) WINE_TRY_CFLAGS([-gdwarf-2])
<br>
+ -g) WINE_TRY_CFLAGS([-g${DEBUGFORMAT:-dwarf-2}])
<br>
WINE_TRY_CFLAGS([-gstrict-dwarf]) ;;
<br>
esac
<br>
done
<br>
</blockquote>
<br>
<br>
Maybe we could make it a bit smarter, by scanning CFLAGS for
explicit -gdwarf-X like CROSSDEBUG does. That should in fact make
exposing DEBUGFORMAT redundant. In case of CROSSDEBUG, we also
have split debug option that somehow needs to be passed to
configure, but we don't have that for native parts. (Still,
DEBUGFORMAT may be useful for convenience).
<br>
<br>
<br>
Thanks,
<br>
<br>
Jacek
<br>
<br>
<br>
</blockquote>
<p><font face="Helvetica, Arial, sans-serif">Hi Jacek,</font></p>
<p><font face="Helvetica, Arial, sans-serif">yes, that would be more
coherent</font></p>
<p><font face="Helvetica, Arial, sans-serif"><br>
</font></p>
<p><font face="Helvetica, Arial, sans-serif">going this path of
coherency, would also require to have consistent naming between
CROSSDEBUG and DEBUGFORMAT</font></p>
<p>I found the simple DEBUG a bit misleading, hence the DEBUGFORMAT
choice</p>
<p>so renaming CROSSDEBUG into CROSSDEBUGFORMAT (or supporting both
CROSSDEBUG and CROSSDEBUGFORMAT for easing the transition) would
be even cleaner IMO</p>
<p>I'll resend along those lines</p>
<p>A+<br>
</p>
</body>
</html>