:) 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...

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?

Regards, Markus

-----Ursprüngliche Nachricht----- Von: Johan Engelen [mailto:jbc.engelen@...2592...] Gesendet: Sonntag, 23. März 2014 14:42 An: Markus Engel; 'Inkscape-Devel' Betreff: Re: AW: [Inkscape-devel] Possibly unitialized used of variable

On 23-3-2014 14:05, Markus Engel wrote:

Hi, neither gcc 4.8.1 nor clang 3.4 nor msvc 12 produce this warning, so I think this might be a false positive that has been fixed in newer compiler versions. I don't see how _b[1] could be not initialized...

Thanks for checking with newer compilers.

Here's the warning I currently like best: ../../trunk/src/gradient-drag.cpp:2390:69: warning: variable 'p' is uninitialized when used within its own initialization [-Wuninitialized] Geom::Point p = ls.pointAt(ls.nearestPoint(dragger->point + p));

My goodness, what a terrible bug. Fixed in rev. 13186.

Does everybody now see why warnings are important?! Yes newer compilers may throw more warnings, but they are *better* warnings and we *need* to error on them, *by default*.

Thanks, Johan

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

Ease of use / laziness. It would help if I could share diskspace from host to VM, so only the compilation is done on the VM, but I still have the file handling / bzr / etc. from Windows. Haven't looked into how to do that.

Use Vagrant. (http://www.vagrantup.com/)

I tried it out with Inkscape in January primarily for GUI testing but it's a better fit for your compilation problem [1]. Your exact use case is what I use during my master thesis project. I develop Internet of Things related software in Ada (on my host) and let a VM (Virtualbox) be the build machine. This allows me to separate specific build tools to the VM with ease. The Vagrant settings allows you to map host-folders to VM-folders in a one line configuration.

I haven't tried Windows as host or guest yet though but could be interesting to try out.

[1] https://code.launchpad.net/~inkscape.dev/inkscape/vagrant-no-desktop

Regards

The buildbots on Ubuntu produce a warning (e.g. Saucy, GCC 4.8.2) https://launchpadlibrarian.net/170132517/buildlog_ubuntu-saucy-amd64.inkscape-trunk_1%3A0.48%2Bdevel%2B13168%2B47~ubuntu13.10.1_UPLOADING.txt.gz

In file included from ../../src/2geom/interval.h:45:0, from ../../src/2geom/rect.h:45, from ../../src/display/sp-canvas.cpp:26: ../../src/2geom/generic-interval.h: In static member function 'static void SPCanvasGroup::update(SPCanvasItem*, const Geom::Affine&, unsigned int)': ../../src/2geom/generic-interval.h:144:8: warning: '*((void*)(& bounds)+24).Geom::GenericInterval<double>::_b[1]' may be used uninitialized in this function [-Wmaybe-uninitialized] if(val > _b[1]) _b[1] = val; //no else, as we want to handle NaN ^ ../../src/display/sp-canvas.cpp:1051:19: note: '*((void*)(& bounds)+24).Geom::GenericInterval<double>::_b[1]' was declared here Geom::OptRect bounds; ^ In file included from ../../src/2geom/interval.h:45:0, from ../../src/2geom/rect.h:45, from ../../src/display/sp-canvas.cpp:26: ../../src/2geom/generic-interval.h:143:8: warning: '*((void*)(& bounds)+24).Geom::GenericInterval<double>::_b[0]' may be used uninitialized in this function [-Wmaybe-uninitialized] if(val < _b[0]) _b[0] = val; ^ ../../src/display/sp-canvas.cpp:1051:19: note: '*((void*)(& bounds)+24).Geom::GenericInterval<double>::_b[0]' was declared here Geom::OptRect bounds; ^ In file included from ../../src/2geom/interval.h:45:0, from ../../src/2geom/rect.h:45, from ../../src/display/sp-canvas.cpp:26: ../../src/2geom/generic-interval.h:144:8: warning: '*((void*)(& bounds)+8).Geom::GenericInterval<double>::_b[1]' may be used uninitialized in this function [-Wmaybe-uninitialized] if(val > _b[1]) _b[1] = val; //no else, as we want to handle NaN ^ ../../src/display/sp-canvas.cpp:1051:19: note: '*((void*)(& bounds)+8).Geom::GenericInterval<double>::_b[1]' was declared here Geom::OptRect bounds; ^ In file included from ../../src/2geom/interval.h:45:0, from ../../src/2geom/rect.h:45, from ../../src/display/sp-canvas.cpp:26: ../../src/2geom/generic-interval.h:143:8: warning: '*((void*)(& bounds)+8).Geom::GenericInterval<double>::_b[0]' may be used uninitialized in this function [-Wmaybe-uninitialized] if(val < _b[0]) _b[0] = val; ^ ../../src/display/sp-canvas.cpp:1051:19: note: '*((void*)(& bounds)+8).Geom::GenericInterval<double>::_b[0]' was declared here Geom::OptRect bounds; ^

On 2014-03-23 14:05 +0100, Markus Engel wrote:

Hi, neither gcc 4.8.1 nor clang 3.4 nor msvc 12 produce this warning, so I think this might be a false positive that has been fixed in newer compiler versions. I don't see how _b[1] could be not initialized...

Here's the warning I currently like best: ../../trunk/src/gradient-drag.cpp:2390:69: warning: variable 'p' is uninitialized when used within its own initialization [-Wuninitialized] Geom::Point p = ls.pointAt(ls.nearestPoint(dragger->point + p));

Regards, Markus

-----Ursprüngliche Nachricht----- Von: Johan Engelen [mailto:jbc.engelen@...2592...] Gesendet: Sonntag, 23. März 2014 13:24 An: Inkscape-Devel Betreff: [Inkscape-devel] Possibly unitialized used of variable

Hi all, GCC 4.6.1 gives an uninit var warning in sp-canvas.cpp:

src/2geom/generic-interval.h: In static member function 'static void SPCanvasGroup::update(SPCanvasItem*, const Geom::Affine&, unsigned int)':

src/2geom/generic-interval.h:144:8: error: '*((void*)(& bounds)+24).Geom::GenericInterval<double>::_b[1]' may be used uninitialized in this function [-Werror=uninitialized]

src/display/sp-canvas.cpp:1051:19: note: '*((void*)(& bounds)+24).Geom::GenericInterval<double>::_b[1]' was declared here

...

I have been working on this warning for hours, but I cannot find the problem. Because I have already found a (serious) bug from another warning like this, I am very reluctant to conclude that it is a compiler mis-analysis.

I would greatly appreciate someone else having a (thorough) look at this.

Thanks, Johan

---

-- Learn Graph Databases - Download FREE O'Reilly Book "Graph Databases" is the definitive new guide to graph databases and their applications. Written by three acclaimed leaders in the field, this first edition is now available. Download your free book today! http://p.sf.net/sfu/13534_NeoTech _______________________________________________ Inkscape-devel mailing list Inkscape-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/inkscape-devel

---

Learn Graph Databases - Download FREE O'Reilly Book "Graph Databases" is the definitive new guide to graph databases and their applications. Written by three acclaimed leaders in the field, this first edition is now available. Download your free book today! http://p.sf.net/sfu/13534_NeoTech _______________________________________________ Inkscape-devel mailing list Inkscape-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/inkscape-devel