Why don't you use a vm? Later we should make sure that we let configure automatically choose warning options depending on the used compiler: warning: unknown warning option '-Wno-unused-but-set-variable'; did you mean '-Wno-unused-variable'? [-Wunknown-warning-option]
../../trunk/src/gradient-drag.cpp:1491:26: warning: comparison of constant -1 with expression of type 'GrPointType' is always false [-Wtautological-constant-out-of-range-compare] (point_type == -1 || da->point_type == point_type) && ----- ../../trunk/src/extension/internal/emf-print.cpp:948:49: warning: variable 'i' is uninitialized when used here [-Wuninitialized] for(int i; cit != pathRect.end_open();++cit,i++) { ^ ../../trunk/src/extension/internal/emf-print.cpp:948:14: note: initialize the variable 'i' to silence this warning for(int i; cit != pathRect.end_open();++cit,i++) { ^ = 0
I could provide you with a list of warnings emitted by clang, if you want me to :) .
Regards, Markus
-----Ursprüngliche Nachricht----- Von: Johan Engelen [mailto:jbc.engelen@...2592...] Gesendet: Sonntag, 23. März 2014 15:07 An: Markus Engel; 'Inkscape-Devel' Betreff: Re: AW: AW: [Inkscape-devel] Possibly unitialized used of variable
On 23-3-2014 14:51, Markus Engel wrote:
:) Actually, clang has already output this warning a year ago, but I never managed to look into this... gcc did never complain. You should probably get a few different environments (clang with -Weverything ;)) so you can fix even more warnings...
I so desperately want to build with clang on Windows... :'( :'( :'(
In many places I'm just not sure whether these things are intended. Let's say ../../trunk/src/live_effects/lpe-powerstroke.cpp:611:104: warning: implicit conversion from 'double' to 'bool' changes value from 0.01 to
false
[-Wliteral-conversion] Geom::Path fixed_path = path_from_piecewise_fix_cusps(
pwd2_out,
y, jointype, miter_limit, LPE_CONVERSION_TOLERANCE);
#define LPE_CONVERSION_TOLERANCE 0.01 // FIXME: find good solution for this.
The last parameter is a bool though. So?
Again you found a terrible bug! I know my powerstroke code needs a LOT of love, but this is really a bad coding bug. The bug here was that the caller things it is setting the tolerance but instead it is setting a bool parameter that is unused, and the conversion tolerance is set to the default value.
See the fix in rev. 13188.
So this was not a false positive at all.
ciao, Johan