A bezier curve found in an application has caused endless recursive calls to flatten_bezier due to float point precision limitations in the flatness test. This curve has been added to the graphicpath.c test to check for the resulting stack overflow error.
Wine-Bug: https://bugs.winehq.org/show_bug.cgi?id=52492 Signed-off-by: Robert Feuerbach rjfeuerbach@gmail.com --- dlls/gdiplus/tests/graphicspath.c | 55 ++++++++++++++++++++++++++++++- 1 file changed, 54 insertions(+), 1 deletion(-)
diff --git a/dlls/gdiplus/tests/graphicspath.c b/dlls/gdiplus/tests/graphicspath.c index 07e6dd166ad..36508776ed3 100644 --- a/dlls/gdiplus/tests/graphicspath.c +++ b/dlls/gdiplus/tests/graphicspath.c @@ -1253,6 +1253,58 @@ static void test_flatten(void) GdipDeletePath(path); }
+static void test_flatten2(void) +{ + GpStatus status; + GpPath *path; + INT count; + + status = GdipCreatePath(0, &path); + expect(Ok, status); + status = GdipStartPathFigure(path); + expect(Ok, status); + /* path seen in the wild that caused a stack overflow */ + /* low-precision points that can cause a crash */ + status = GdipAddPathBezier(path, + 154.950806, 33.391144, + 221.586075, 15.536285, + 291.747314, 15.536285, + 358.382568, 33.391144); + expect(Ok, status); + status = GdipAddPathBezier(path, + 256.666809, 412.999512, + 256.666718, 412.999481, + 256.666656, 412.999481, + 256.666565, 412.999512); + expect(Ok, status); + status = GdipClosePathFigure(path); + expect(Ok, status); + trace("Flattening good and problematic bezier curves\n"); + status = GdipFlattenPath(path, NULL, 1.0); + + /* now redo with hexadecimal floats to replicate the points exactly */ + status = GdipResetPath(path); + expect(Ok, status); + status = GdipAddPathBezier(path, + 0x1.35e6d00000000p+7, 0x1.0b21100000000p+5, + 0x1.bb2c120000000p+7, 0x1.f129400000000p+3, + 0x1.23bf500000000p+8, 0x1.f129400000000p+3, + 0x1.6661f00000000p+8, 0x1.0b21100000000p+5); + expect(Ok, status); + status = GdipAddPathBezier(path, + 0x1.00aab40000000p+8, 0x1.9cffe00000000p+8, + 0x1.00aaae0000000p+8, 0x1.9cffde0000000p+8, + 0x1.00aaaa0000000p+8, 0x1.9cffde0000000p+8, + 0x1.00aaa40000000p+8, 0x1.9cffe00000000p+8); + expect(Ok, status); + status = GdipClosePathFigure(path); + expect(Ok, status); + trace("Flattening precise copied problematic curves\n"); + status = GdipFlattenPath(path, NULL, 1.0); + + GdipDeletePath(path); +} + static path_test_t widenline_path[] = { {5.0, 5.0, PathPointTypeStart, 0, 0}, /*0*/ {50.0, 5.0, PathPointTypeLine, 0, 0}, /*1*/ @@ -1900,6 +1952,7 @@ START_TEST(graphicspath) test_widen_cap(); test_isvisible(); test_empty_rect(); - + test_flatten2(); + GdiplusShutdown(gdiplusToken); }
The float equality and flatness calculation in flatten_bezier can fail due to the limited precision of the float math. The equality test was replaced with a simple check against the given flatness tolerance.
Wine-Bug: https://bugs.winehq.org/show_bug.cgi?id=52492 Signed-off-by: Robert Feuerbach rjfeuerbach@gmail.com --- dlls/gdiplus/graphicspath.c | 11 ++++++----- 1 file changed, 6 insertions(+), 5 deletions(-)
diff --git a/dlls/gdiplus/graphicspath.c b/dlls/gdiplus/graphicspath.c index ce2666eedab..c1bdf226e52 100644 --- a/dlls/gdiplus/graphicspath.c +++ b/dlls/gdiplus/graphicspath.c @@ -109,7 +109,7 @@ static INT path_list_count(path_list_node_t *node) * - (x2, y2): first control point; * - (x3, y3): second control point; * - end : pointer to end point node - * - flatness: admissible error of linear approximation. + * - flatness: admissible error of linear approximation in coordinate units. * * Return value: * TRUE : success @@ -142,12 +142,13 @@ static BOOL flatten_bezier(path_list_node_t *start, REAL x2, REAL y2, REAL x3, R mp[2].X = (mp[1].X + mp[3].X) / 2.0; mp[2].Y = (mp[1].Y + mp[3].Y) / 2.0;
- if ((x2 == mp[0].X && y2 == mp[0].Y && x3 == mp[1].X && y3 == mp[1].Y) || - (x2 == mp[3].X && y2 == mp[3].Y && x3 == mp[4].X && y3 == mp[4].Y)) - return TRUE; - pt = end->pt; pt_st = start->pt; + /* test for closely spaced points to avoid limited-precision errors in flatness check */ + if((fabs(pt.X - mp[2].X) + fabs(pt.Y - mp[2].Y) + + fabs(pt_st.X - mp[2].X) + fabs(pt_st.Y - mp[2].Y) ) <= flatness) + return TRUE; + /* check flatness as a half of distance between middle point and a linearized path */ if(fabs(((pt.Y - pt_st.Y)*mp[2].X + (pt_st.X - pt.X)*mp[2].Y + (pt_st.Y*pt.X - pt_st.X*pt.Y))) <=
More of a general question towards the wine devs: According to gcc we have a lot of such float comparisons in the code. Isn't that a potential general problem?
Regards, Fabian Maurer
On Freitag, 4. Februar 2022 16:58:48 CET Robert Feuerbach wrote:
The float equality and flatness calculation in flatten_bezier can fail due to the limited precision of the float math. The equality test was replaced with a simple check against the given flatness tolerance.
Wine-Bug: https://bugs.winehq.org/show_bug.cgi?id=52492 Signed-off-by: Robert Feuerbach rjfeuerbach@gmail.com
dlls/gdiplus/graphicspath.c | 11 ++++++----- 1 file changed, 6 insertions(+), 5 deletions(-)
diff --git a/dlls/gdiplus/graphicspath.c b/dlls/gdiplus/graphicspath.c index ce2666eedab..c1bdf226e52 100644 --- a/dlls/gdiplus/graphicspath.c +++ b/dlls/gdiplus/graphicspath.c @@ -109,7 +109,7 @@ static INT path_list_count(path_list_node_t *node)
- (x2, y2): first control point;
- (x3, y3): second control point;
- end : pointer to end point node
- flatness: admissible error of linear approximation.
- flatness: admissible error of linear approximation in coordinate
units. *
- Return value:
- TRUE : success
@@ -142,12 +142,13 @@ static BOOL flatten_bezier(path_list_node_t *start, REAL x2, REAL y2, REAL x3, R mp[2].X = (mp[1].X + mp[3].X) / 2.0; mp[2].Y = (mp[1].Y + mp[3].Y) / 2.0;
- if ((x2 == mp[0].X && y2 == mp[0].Y && x3 == mp[1].X && y3 == mp[1].Y)
|| - (x2 == mp[3].X && y2 == mp[3].Y && x3 == mp[4].X && y3 == mp[4].Y)) - return TRUE;
- pt = end->pt; pt_st = start->pt;
- /* test for closely spaced points to avoid limited-precision errors in
flatness check */ + if((fabs(pt.X - mp[2].X) + fabs(pt.Y - mp[2].Y) +
fabs(pt_st.X - mp[2].X) + fabs(pt_st.Y - mp[2].Y) ) <= flatness)return TRUE;- /* check flatness as a half of distance between middle point and a
linearized path */ if(fabs(((pt.Y - pt_st.Y)*mp[2].X + (pt_st.X - pt.X)*mp[2].Y + (pt_st.Y*pt.X - pt_st.X*pt.Y))) <=
It's possible, but I'd say there are also correct ways to use these comparisons.
On Fri, Feb 4, 2022 at 10:03 AM Fabian Maurer dark.shadow4@web.de wrote:
More of a general question towards the wine devs: According to gcc we have a lot of such float comparisons in the code. Isn't that a potential general problem?
Regards, Fabian Maurer
On Freitag, 4. Februar 2022 16:58:48 CET Robert Feuerbach wrote:
The float equality and flatness calculation in flatten_bezier can fail due to the limited precision of the float math. The equality test was replaced with a simple check against the given flatness tolerance.
Wine-Bug: https://bugs.winehq.org/show_bug.cgi?id=52492 Signed-off-by: Robert Feuerbach rjfeuerbach@gmail.com
dlls/gdiplus/graphicspath.c | 11 ++++++----- 1 file changed, 6 insertions(+), 5 deletions(-)
diff --git a/dlls/gdiplus/graphicspath.c b/dlls/gdiplus/graphicspath.c index ce2666eedab..c1bdf226e52 100644 --- a/dlls/gdiplus/graphicspath.c +++ b/dlls/gdiplus/graphicspath.c @@ -109,7 +109,7 @@ static INT path_list_count(path_list_node_t *node)
- (x2, y2): first control point;
- (x3, y3): second control point;
- end : pointer to end point node
- flatness: admissible error of linear approximation.
- flatness: admissible error of linear approximation in coordinate
units. *
- Return value:
- TRUE : success
@@ -142,12 +142,13 @@ static BOOL flatten_bezier(path_list_node_t *start, REAL x2, REAL y2, REAL x3, R mp[2].X = (mp[1].X + mp[3].X) / 2.0; mp[2].Y = (mp[1].Y + mp[3].Y) / 2.0;
- if ((x2 == mp[0].X && y2 == mp[0].Y && x3 == mp[1].X && y3 == mp[1].Y)
|| - (x2 == mp[3].X && y2 == mp[3].Y && x3 == mp[4].X && y3 == mp[4].Y)) - return TRUE;
- pt = end->pt; pt_st = start->pt;
- /* test for closely spaced points to avoid limited-precision errors in
flatness check */ + if((fabs(pt.X - mp[2].X) + fabs(pt.Y - mp[2].Y) +
fabs(pt_st.X - mp[2].X) + fabs(pt_st.Y - mp[2].Y) ) <= flatness)return TRUE;- /* check flatness as a half of distance between middle point and a
linearized path */ if(fabs(((pt.Y - pt_st.Y)*mp[2].X + (pt_st.X - pt.X)*mp[2].Y + (pt_st.Y*pt.X - pt_st.X*pt.Y))) <=
I'm not convinced by this approach. There might be a situation where the midpoint between the endpoints happens to be close to the midpoint between the control points, but the control points are further than the flatness, leaving a possibility that we stop while our approximation has more error than the flatness.
I feel like we should be testing the distance between the control points and the line segment (or the distance between the control points and an endpoint, which should be no less than the distance from the line segment).
I don't really understand the existing "half of distance between middle point and a linearized path" test either. Nikolay, can you explain this?
On Fri, Feb 4, 2022 at 9:59 AM Robert Feuerbach rjfeuerbach@gmail.com wrote:
The float equality and flatness calculation in flatten_bezier can fail due to the limited precision of the float math. The equality test was replaced with a simple check against the given flatness tolerance.
Wine-Bug: https://bugs.winehq.org/show_bug.cgi?id=52492 Signed-off-by: Robert Feuerbach rjfeuerbach@gmail.com
dlls/gdiplus/graphicspath.c | 11 ++++++----- 1 file changed, 6 insertions(+), 5 deletions(-)
diff --git a/dlls/gdiplus/graphicspath.c b/dlls/gdiplus/graphicspath.c index ce2666eedab..c1bdf226e52 100644 --- a/dlls/gdiplus/graphicspath.c +++ b/dlls/gdiplus/graphicspath.c @@ -109,7 +109,7 @@ static INT path_list_count(path_list_node_t *node)
- (x2, y2): first control point;
- (x3, y3): second control point;
- end : pointer to end point node
- flatness: admissible error of linear approximation.
- flatness: admissible error of linear approximation in coordinate units.
- Return value:
- TRUE : success
@@ -142,12 +142,13 @@ static BOOL flatten_bezier(path_list_node_t *start, REAL x2, REAL y2, REAL x3, R mp[2].X = (mp[1].X + mp[3].X) / 2.0; mp[2].Y = (mp[1].Y + mp[3].Y) / 2.0;
- if ((x2 == mp[0].X && y2 == mp[0].Y && x3 == mp[1].X && y3 == mp[1].Y) ||
(x2 == mp[3].X && y2 == mp[3].Y && x3 == mp[4].X && y3 == mp[4].Y))return TRUE;- pt = end->pt; pt_st = start->pt;
- /* test for closely spaced points to avoid limited-precision errors in flatness check */
- if((fabs(pt.X - mp[2].X) + fabs(pt.Y - mp[2].Y) +
fabs(pt_st.X - mp[2].X) + fabs(pt_st.Y - mp[2].Y) ) <= flatness)return TRUE;- /* check flatness as a half of distance between middle point and a linearized path */ if(fabs(((pt.Y - pt_st.Y)*mp[2].X + (pt_st.X - pt.X)*mp[2].Y + (pt_st.Y*pt.X - pt_st.X*pt.Y))) <=
-- 2.32.0 (Apple Git-132)
On 2/4/22 20:33, Esme Povirk (she/they) wrote:
I'm not convinced by this approach. There might be a situation where the midpoint between the endpoints happens to be close to the midpoint between the control points, but the control points are further than the flatness, leaving a possibility that we stop while our approximation has more error than the flatness.
I feel like we should be testing the distance between the control points and the line segment (or the distance between the control points and an endpoint, which should be no less than the distance from the line segment).
I don't really understand the existing "half of distance between middle point and a linearized path" test either. Nikolay, can you explain this?
It's be a while. The idea is to check a distance from mp[2] point to {pt, pt_st} segment (dist_m), and stop recursion if it's small enough.
I haven't verified right now, but expression I believe was meant to get that distance as:
sqr(dot(mp - pt, pt_st - pt)) + sqr(dist_m) = sqr(dist(pt_st, pt));
On Fri, Feb 4, 2022 at 9:59 AM Robert Feuerbach rjfeuerbach@gmail.com wrote:
The float equality and flatness calculation in flatten_bezier can fail due to the limited precision of the float math. The equality test was replaced with a simple check against the given flatness tolerance.
Wine-Bug: https://bugs.winehq.org/show_bug.cgi?id=52492 Signed-off-by: Robert Feuerbach rjfeuerbach@gmail.com
dlls/gdiplus/graphicspath.c | 11 ++++++----- 1 file changed, 6 insertions(+), 5 deletions(-)
diff --git a/dlls/gdiplus/graphicspath.c b/dlls/gdiplus/graphicspath.c index ce2666eedab..c1bdf226e52 100644 --- a/dlls/gdiplus/graphicspath.c +++ b/dlls/gdiplus/graphicspath.c @@ -109,7 +109,7 @@ static INT path_list_count(path_list_node_t *node)
- (x2, y2): first control point;
- (x3, y3): second control point;
- end : pointer to end point node
- flatness: admissible error of linear approximation.
- flatness: admissible error of linear approximation in coordinate units.
- Return value:
- TRUE : success
@@ -142,12 +142,13 @@ static BOOL flatten_bezier(path_list_node_t *start, REAL x2, REAL y2, REAL x3, R mp[2].X = (mp[1].X + mp[3].X) / 2.0; mp[2].Y = (mp[1].Y + mp[3].Y) / 2.0;
- if ((x2 == mp[0].X && y2 == mp[0].Y && x3 == mp[1].X && y3 == mp[1].Y) ||
(x2 == mp[3].X && y2 == mp[3].Y && x3 == mp[4].X && y3 == mp[4].Y))return TRUE;pt = end->pt; pt_st = start->pt;
- /* test for closely spaced points to avoid limited-precision errors in flatness check */
- if((fabs(pt.X - mp[2].X) + fabs(pt.Y - mp[2].Y) +
fabs(pt_st.X - mp[2].X) + fabs(pt_st.Y - mp[2].Y) ) <= flatness)return TRUE;/* check flatness as a half of distance between middle point and a linearized path */ if(fabs(((pt.Y - pt_st.Y)*mp[2].X + (pt_st.X - pt.X)*mp[2].Y + (pt_st.Y*pt.X - pt_st.X*pt.Y))) <=-- 2.32.0 (Apple Git-132)
Couldn't that point be closer to the line segment than the actual curve, if the control points are on opposite sides of the line?
On Fri, Feb 4, 2022 at 1:12 PM Nikolay Sivov nsivov@codeweavers.com wrote:
On 2/4/22 20:33, Esme Povirk (she/they) wrote:
I'm not convinced by this approach. There might be a situation where the midpoint between the endpoints happens to be close to the midpoint between the control points, but the control points are further than the flatness, leaving a possibility that we stop while our approximation has more error than the flatness.
I feel like we should be testing the distance between the control points and the line segment (or the distance between the control points and an endpoint, which should be no less than the distance from the line segment).
I don't really understand the existing "half of distance between middle point and a linearized path" test either. Nikolay, can you explain this?
It's be a while. The idea is to check a distance from mp[2] point to {pt, pt_st} segment (dist_m), and stop recursion if it's small enough.
I haven't verified right now, but expression I believe was meant to get that distance as:
sqr(dot(mp - pt, pt_st - pt)) + sqr(dist_m) = sqr(dist(pt_st, pt));
On Fri, Feb 4, 2022 at 9:59 AM Robert Feuerbach rjfeuerbach@gmail.com wrote:
The float equality and flatness calculation in flatten_bezier can fail due to the limited precision of the float math. The equality test was replaced with a simple check against the given flatness tolerance.
Wine-Bug: https://bugs.winehq.org/show_bug.cgi?id=52492 Signed-off-by: Robert Feuerbach rjfeuerbach@gmail.com
dlls/gdiplus/graphicspath.c | 11 ++++++----- 1 file changed, 6 insertions(+), 5 deletions(-)
diff --git a/dlls/gdiplus/graphicspath.c b/dlls/gdiplus/graphicspath.c index ce2666eedab..c1bdf226e52 100644 --- a/dlls/gdiplus/graphicspath.c +++ b/dlls/gdiplus/graphicspath.c @@ -109,7 +109,7 @@ static INT path_list_count(path_list_node_t *node)
- (x2, y2): first control point;
- (x3, y3): second control point;
- end : pointer to end point node
- flatness: admissible error of linear approximation.
- flatness: admissible error of linear approximation in coordinate units.
- Return value:
- TRUE : success
@@ -142,12 +142,13 @@ static BOOL flatten_bezier(path_list_node_t *start, REAL x2, REAL y2, REAL x3, R mp[2].X = (mp[1].X + mp[3].X) / 2.0; mp[2].Y = (mp[1].Y + mp[3].Y) / 2.0;
- if ((x2 == mp[0].X && y2 == mp[0].Y && x3 == mp[1].X && y3 == mp[1].Y) ||
(x2 == mp[3].X && y2 == mp[3].Y && x3 == mp[4].X && y3 == mp[4].Y))return TRUE;pt = end->pt; pt_st = start->pt;
- /* test for closely spaced points to avoid limited-precision errors in flatness check */
- if((fabs(pt.X - mp[2].X) + fabs(pt.Y - mp[2].Y) +
fabs(pt_st.X - mp[2].X) + fabs(pt_st.Y - mp[2].Y) ) <= flatness)return TRUE;/* check flatness as a half of distance between middle point and a linearized path */ if(fabs(((pt.Y - pt_st.Y)*mp[2].X + (pt_st.X - pt.X)*mp[2].Y + (pt_st.Y*pt.X - pt_st.X*pt.Y))) <=-- 2.32.0 (Apple Git-132)
A different way to perform the flatten is with a direct approach, instead of recursive, which eliminates the stack overflow risk and also is rather straightforward. The open access reference, "Computer Aided Geometric Design" by Sederberg ( https://scholarsarchive.byu.edu/cgi/viewcontent.cgi?article=1000&context...) has a description that is easily implemented, and I can make a v2 patch with it instead.
I'm also not familiar with the etiquette of this mailing list -- should this type of discussion be emailed about off the list?
On Fri, Feb 4, 2022 at 2:28 PM Esme Povirk (she/they) esme@codeweavers.com wrote:
Couldn't that point be closer to the line segment than the actual curve, if the control points are on opposite sides of the line?
On Fri, Feb 4, 2022 at 1:12 PM Nikolay Sivov nsivov@codeweavers.com wrote:
On 2/4/22 20:33, Esme Povirk (she/they) wrote:
I'm not convinced by this approach. There might be a situation where the midpoint between the endpoints happens to be close to the midpoint between the control points, but the control points are further than the flatness, leaving a possibility that we stop while our approximation has more error than the flatness.
I feel like we should be testing the distance between the control points and the line segment (or the distance between the control points and an endpoint, which should be no less than the distance from the line segment).
I don't really understand the existing "half of distance between middle point and a linearized path" test either. Nikolay, can you explain this?
It's be a while. The idea is to check a distance from mp[2] point to {pt, pt_st} segment (dist_m), and stop recursion if it's small enough.
I haven't verified right now, but expression I believe was meant to get that distance as:
sqr(dot(mp - pt, pt_st - pt)) + sqr(dist_m) = sqr(dist(pt_st, pt));
On Fri, Feb 4, 2022 at 9:59 AM Robert Feuerbach rjfeuerbach@gmail.com
wrote:
The float equality and flatness calculation in flatten_bezier can fail due to the limited precision of the float math. The equality test was replaced with a simple check against the given flatness tolerance.
Wine-Bug: https://bugs.winehq.org/show_bug.cgi?id=52492 Signed-off-by: Robert Feuerbach rjfeuerbach@gmail.com
dlls/gdiplus/graphicspath.c | 11 ++++++----- 1 file changed, 6 insertions(+), 5 deletions(-)
diff --git a/dlls/gdiplus/graphicspath.c b/dlls/gdiplus/graphicspath.c index ce2666eedab..c1bdf226e52 100644 --- a/dlls/gdiplus/graphicspath.c +++ b/dlls/gdiplus/graphicspath.c @@ -109,7 +109,7 @@ static INT path_list_count(path_list_node_t *node)
- (x2, y2): first control point;
- (x3, y3): second control point;
- end : pointer to end point node
- flatness: admissible error of linear approximation.
- flatness: admissible error of linear approximation in
coordinate units.
- Return value:
- TRUE : success
@@ -142,12 +142,13 @@ static BOOL flatten_bezier(path_list_node_t
*start, REAL x2, REAL y2, REAL x3, R
mp[2].X = (mp[1].X + mp[3].X) / 2.0; mp[2].Y = (mp[1].Y + mp[3].Y) / 2.0;
- if ((x2 == mp[0].X && y2 == mp[0].Y && x3 == mp[1].X && y3 ==
mp[1].Y) ||
(x2 == mp[3].X && y2 == mp[3].Y && x3 == mp[4].X && y3 ==mp[4].Y))
return TRUE;pt = end->pt; pt_st = start->pt;
- /* test for closely spaced points to avoid limited-precision
errors in flatness check */
- if((fabs(pt.X - mp[2].X) + fabs(pt.Y - mp[2].Y) +
fabs(pt_st.X - mp[2].X) + fabs(pt_st.Y - mp[2].Y) ) <=flatness)
return TRUE;/* check flatness as a half of distance between middle pointand a linearized path */
if(fabs(((pt.Y - pt_st.Y)*mp[2].X + (pt_st.X - pt.X)*mp[2].Y + (pt_st.Y*pt.X - pt_st.X*pt.Y))) <=-- 2.32.0 (Apple Git-132)
Unless there's a specific reason to discuss it in private, we should do it on the mailing list where everyone can participate.
I consider myself weak in this kind of math, so I'd appreciate having others chime in who may have a better understanding of it.
So if I understand the concept correctly, this would give us a way to calculate the number of line segments we need to approximate the Bezier curve with a specific upper bound on the error, and calculate the approximation without splitting it into more Bezier curves.
This would also enable an alternative approach where we use this method to determine the error of 1 single line segment, and split the curve similar to how we do it now if the error is too large. I don't think we technically needed recursion to begin with because we can insert additional points on the path we're working on. But if the number of points we need can be easily calculated, I don't think there's a clear advantage to this approach.
A possible obstacle is that we have tests for GdipFlattenPath which look for specific points and point counts. One of them still has some todo's meaning we don't quite match native. But, there's a risk that you would do this work and find out that your approach cannot pass our existing tests. If the approach is still bisecting, that probably won't make the situation any worse. (Then again, for all I know, native might do a direct approximation.)
If you think you can make the direct approach work while still passing the tests, I would appreciate that. Regardless, I think it'd be acceptable to keep the recursive approach, but fix the flatness test.
On Mon, Feb 7, 2022 at 12:36 PM Robert Feuerbach rjfeuerbach@gmail.com wrote:
A different way to perform the flatten is with a direct approach, instead of recursive, which eliminates the stack overflow risk and also is rather straightforward. The open access reference, "Computer Aided Geometric Design" by Sederberg (https://scholarsarchive.byu.edu/cgi/viewcontent.cgi?article=1000&context...) has a description that is easily implemented, and I can make a v2 patch with it instead.
I'm also not familiar with the etiquette of this mailing list -- should this type of discussion be emailed about off the list?
On Fri, Feb 4, 2022 at 2:28 PM Esme Povirk (she/they) esme@codeweavers.com wrote:
Couldn't that point be closer to the line segment than the actual curve, if the control points are on opposite sides of the line?
On Fri, Feb 4, 2022 at 1:12 PM Nikolay Sivov nsivov@codeweavers.com wrote:
On 2/4/22 20:33, Esme Povirk (she/they) wrote:
I'm not convinced by this approach. There might be a situation where the midpoint between the endpoints happens to be close to the midpoint between the control points, but the control points are further than the flatness, leaving a possibility that we stop while our approximation has more error than the flatness.
I feel like we should be testing the distance between the control points and the line segment (or the distance between the control points and an endpoint, which should be no less than the distance from the line segment).
I don't really understand the existing "half of distance between middle point and a linearized path" test either. Nikolay, can you explain this?
It's be a while. The idea is to check a distance from mp[2] point to {pt, pt_st} segment (dist_m), and stop recursion if it's small enough.
I haven't verified right now, but expression I believe was meant to get that distance as:
sqr(dot(mp - pt, pt_st - pt)) + sqr(dist_m) = sqr(dist(pt_st, pt));
On Fri, Feb 4, 2022 at 9:59 AM Robert Feuerbach rjfeuerbach@gmail.com wrote:
The float equality and flatness calculation in flatten_bezier can fail due to the limited precision of the float math. The equality test was replaced with a simple check against the given flatness tolerance.
Wine-Bug: https://bugs.winehq.org/show_bug.cgi?id=52492 Signed-off-by: Robert Feuerbach rjfeuerbach@gmail.com
dlls/gdiplus/graphicspath.c | 11 ++++++----- 1 file changed, 6 insertions(+), 5 deletions(-)
diff --git a/dlls/gdiplus/graphicspath.c b/dlls/gdiplus/graphicspath.c index ce2666eedab..c1bdf226e52 100644 --- a/dlls/gdiplus/graphicspath.c +++ b/dlls/gdiplus/graphicspath.c @@ -109,7 +109,7 @@ static INT path_list_count(path_list_node_t *node)
- (x2, y2): first control point;
- (x3, y3): second control point;
- end : pointer to end point node
- flatness: admissible error of linear approximation.
- flatness: admissible error of linear approximation in coordinate units.
- Return value:
- TRUE : success
@@ -142,12 +142,13 @@ static BOOL flatten_bezier(path_list_node_t *start, REAL x2, REAL y2, REAL x3, R mp[2].X = (mp[1].X + mp[3].X) / 2.0; mp[2].Y = (mp[1].Y + mp[3].Y) / 2.0;
- if ((x2 == mp[0].X && y2 == mp[0].Y && x3 == mp[1].X && y3 == mp[1].Y) ||
(x2 == mp[3].X && y2 == mp[3].Y && x3 == mp[4].X && y3 == mp[4].Y))return TRUE;pt = end->pt; pt_st = start->pt;
- /* test for closely spaced points to avoid limited-precision errors in flatness check */
- if((fabs(pt.X - mp[2].X) + fabs(pt.Y - mp[2].Y) +
fabs(pt_st.X - mp[2].X) + fabs(pt_st.Y - mp[2].Y) ) <= flatness)return TRUE;/* check flatness as a half of distance between middle point and a linearized path */ if(fabs(((pt.Y - pt_st.Y)*mp[2].X + (pt_st.X - pt.X)*mp[2].Y + (pt_st.Y*pt.X - pt_st.X*pt.Y))) <=-- 2.32.0 (Apple Git-132)
I tested the direct approach, and it was as you suspected: it was difficult to pass the GdipFlattenPath tests. The spacing of the points could not be easily made to match or improve upon the tests.
Keeping with the recursion approach, by changing the scaling term in the flatness test from 0.5 to 0.577, we are able to pass all the GdipFlattenPath tests and remove the TODOs. This is approximately 1/sqrt(3), which might be motivated by the polynomial terms, but I do not have that math worked out yet. It also is probably not needed in the quest to simply match the native behavior.
I have made updates to dlls/gdiplus/graphicspath.c and dlls/gdiplus/tests/graphicspath.c to capture these results. Should I submit them as an updated, single patch to address both files? If I do them separately, the test metrics will fail since the todo's will succeed when they are expected to fail.
Thanks.
On Mon, Feb 7, 2022 at 5:03 PM Esme Povirk (she/they) esme@codeweavers.com wrote:
Unless there's a specific reason to discuss it in private, we should do it on the mailing list where everyone can participate.
I consider myself weak in this kind of math, so I'd appreciate having others chime in who may have a better understanding of it.
So if I understand the concept correctly, this would give us a way to calculate the number of line segments we need to approximate the Bezier curve with a specific upper bound on the error, and calculate the approximation without splitting it into more Bezier curves.
This would also enable an alternative approach where we use this method to determine the error of 1 single line segment, and split the curve similar to how we do it now if the error is too large. I don't think we technically needed recursion to begin with because we can insert additional points on the path we're working on. But if the number of points we need can be easily calculated, I don't think there's a clear advantage to this approach.
A possible obstacle is that we have tests for GdipFlattenPath which look for specific points and point counts. One of them still has some todo's meaning we don't quite match native. But, there's a risk that you would do this work and find out that your approach cannot pass our existing tests. If the approach is still bisecting, that probably won't make the situation any worse. (Then again, for all I know, native might do a direct approximation.)
If you think you can make the direct approach work while still passing the tests, I would appreciate that. Regardless, I think it'd be acceptable to keep the recursive approach, but fix the flatness test.
On Mon, Feb 7, 2022 at 12:36 PM Robert Feuerbach rjfeuerbach@gmail.com wrote:
A different way to perform the flatten is with a direct approach,
instead of recursive, which eliminates the stack overflow risk and also is rather straightforward. The open access reference, "Computer Aided Geometric Design" by Sederberg ( https://scholarsarchive.byu.edu/cgi/viewcontent.cgi?article=1000&context...) has a description that is easily implemented, and I can make a v2 patch with it instead.
I'm also not familiar with the etiquette of this mailing list -- should
this type of discussion be emailed about off the list?
On Fri, Feb 4, 2022 at 2:28 PM Esme Povirk (she/they) <
esme@codeweavers.com> wrote:
Couldn't that point be closer to the line segment than the actual curve, if the control points are on opposite sides of the line?
On Fri, Feb 4, 2022 at 1:12 PM Nikolay Sivov nsivov@codeweavers.com
wrote:
On 2/4/22 20:33, Esme Povirk (she/they) wrote:
I'm not convinced by this approach. There might be a situation where the midpoint between the endpoints happens to be close to the
midpoint
between the control points, but the control points are further than the flatness, leaving a possibility that we stop while our approximation has more error than the flatness.
I feel like we should be testing the distance between the control points and the line segment (or the distance between the control points and an endpoint, which should be no less than the distance
from
the line segment).
I don't really understand the existing "half of distance between middle point and a linearized path" test either. Nikolay, can you explain this?
It's be a while. The idea is to check a distance from mp[2] point to {pt, pt_st} segment (dist_m), and stop recursion if it's small enough.
I haven't verified right now, but expression I believe was meant to
get
that distance as:
sqr(dot(mp - pt, pt_st - pt)) + sqr(dist_m) = sqr(dist(pt_st, pt));
On Fri, Feb 4, 2022 at 9:59 AM Robert Feuerbach <
rjfeuerbach@gmail.com> wrote:
The float equality and flatness calculation in flatten_bezier can fail due to the limited precision of the float math. The equality test was replaced with a simple check against the given flatness tolerance.
Wine-Bug: https://bugs.winehq.org/show_bug.cgi?id=52492 Signed-off-by: Robert Feuerbach rjfeuerbach@gmail.com
dlls/gdiplus/graphicspath.c | 11 ++++++----- 1 file changed, 6 insertions(+), 5 deletions(-)
diff --git a/dlls/gdiplus/graphicspath.c
b/dlls/gdiplus/graphicspath.c
index ce2666eedab..c1bdf226e52 100644 --- a/dlls/gdiplus/graphicspath.c +++ b/dlls/gdiplus/graphicspath.c @@ -109,7 +109,7 @@ static INT path_list_count(path_list_node_t
*node)
- (x2, y2): first control point;
- (x3, y3): second control point;
- end : pointer to end point node
- flatness: admissible error of linear approximation.
- flatness: admissible error of linear approximation in
coordinate units.
- Return value:
- TRUE : success
@@ -142,12 +142,13 @@ static BOOL flatten_bezier(path_list_node_t
*start, REAL x2, REAL y2, REAL x3, R
mp[2].X = (mp[1].X + mp[3].X) / 2.0; mp[2].Y = (mp[1].Y + mp[3].Y) / 2.0;
- if ((x2 == mp[0].X && y2 == mp[0].Y && x3 == mp[1].X && y3 ==
mp[1].Y) ||
(x2 == mp[3].X && y2 == mp[3].Y && x3 == mp[4].X && y3 ==mp[4].Y))
return TRUE;pt = end->pt; pt_st = start->pt;
- /* test for closely spaced points to avoid limited-precision
errors in flatness check */
- if((fabs(pt.X - mp[2].X) + fabs(pt.Y - mp[2].Y) +
fabs(pt_st.X - mp[2].X) + fabs(pt_st.Y - mp[2].Y) ) <=flatness)
return TRUE;/* check flatness as a half of distance between middle pointand a linearized path */
if(fabs(((pt.Y - pt_st.Y)*mp[2].X + (pt_st.X - pt.X)*mp[2].Y
(pt_st.Y*pt.X - pt_st.X*pt.Y))) <=-- 2.32.0 (Apple Git-132)
If you fix a test that has a todo, you should always remove the todo_wine in the same patch.
As much as I don't understand why 1/sqrt(3) makes sense, I don't understand why we use 1/2 either.
On Tue, Feb 15, 2022 at 2:51 PM Robert Feuerbach rjfeuerbach@gmail.com wrote:
I tested the direct approach, and it was as you suspected: it was difficult to pass the GdipFlattenPath tests. The spacing of the points could not be easily made to match or improve upon the tests.
Keeping with the recursion approach, by changing the scaling term in the flatness test from 0.5 to 0.577, we are able to pass all the GdipFlattenPath tests and remove the TODOs. This is approximately 1/sqrt(3), which might be motivated by the polynomial terms, but I do not have that math worked out yet. It also is probably not needed in the quest to simply match the native behavior.
I have made updates to dlls/gdiplus/graphicspath.c and dlls/gdiplus/tests/graphicspath.c to capture these results. Should I submit them as an updated, single patch to address both files? If I do them separately, the test metrics will fail since the todo's will succeed when they are expected to fail.
Thanks.
On Mon, Feb 7, 2022 at 5:03 PM Esme Povirk (she/they) esme@codeweavers.com wrote:
Unless there's a specific reason to discuss it in private, we should do it on the mailing list where everyone can participate.
I consider myself weak in this kind of math, so I'd appreciate having others chime in who may have a better understanding of it.
So if I understand the concept correctly, this would give us a way to calculate the number of line segments we need to approximate the Bezier curve with a specific upper bound on the error, and calculate the approximation without splitting it into more Bezier curves.
This would also enable an alternative approach where we use this method to determine the error of 1 single line segment, and split the curve similar to how we do it now if the error is too large. I don't think we technically needed recursion to begin with because we can insert additional points on the path we're working on. But if the number of points we need can be easily calculated, I don't think there's a clear advantage to this approach.
A possible obstacle is that we have tests for GdipFlattenPath which look for specific points and point counts. One of them still has some todo's meaning we don't quite match native. But, there's a risk that you would do this work and find out that your approach cannot pass our existing tests. If the approach is still bisecting, that probably won't make the situation any worse. (Then again, for all I know, native might do a direct approximation.)
If you think you can make the direct approach work while still passing the tests, I would appreciate that. Regardless, I think it'd be acceptable to keep the recursive approach, but fix the flatness test.
On Mon, Feb 7, 2022 at 12:36 PM Robert Feuerbach rjfeuerbach@gmail.com wrote:
A different way to perform the flatten is with a direct approach, instead of recursive, which eliminates the stack overflow risk and also is rather straightforward. The open access reference, "Computer Aided Geometric Design" by Sederberg (https://scholarsarchive.byu.edu/cgi/viewcontent.cgi?article=1000&context...) has a description that is easily implemented, and I can make a v2 patch with it instead.
I'm also not familiar with the etiquette of this mailing list -- should this type of discussion be emailed about off the list?
On Fri, Feb 4, 2022 at 2:28 PM Esme Povirk (she/they) esme@codeweavers.com wrote:
Couldn't that point be closer to the line segment than the actual curve, if the control points are on opposite sides of the line?
On Fri, Feb 4, 2022 at 1:12 PM Nikolay Sivov nsivov@codeweavers.com wrote:
On 2/4/22 20:33, Esme Povirk (she/they) wrote:
I'm not convinced by this approach. There might be a situation where the midpoint between the endpoints happens to be close to the midpoint between the control points, but the control points are further than the flatness, leaving a possibility that we stop while our approximation has more error than the flatness.
I feel like we should be testing the distance between the control points and the line segment (or the distance between the control points and an endpoint, which should be no less than the distance from the line segment).
I don't really understand the existing "half of distance between middle point and a linearized path" test either. Nikolay, can you explain this?
It's be a while. The idea is to check a distance from mp[2] point to {pt, pt_st} segment (dist_m), and stop recursion if it's small enough.
I haven't verified right now, but expression I believe was meant to get that distance as:
sqr(dot(mp - pt, pt_st - pt)) + sqr(dist_m) = sqr(dist(pt_st, pt));
On Fri, Feb 4, 2022 at 9:59 AM Robert Feuerbach rjfeuerbach@gmail.com wrote: > The float equality and flatness calculation in flatten_bezier > can fail due to the limited precision of the float math. > The equality test was replaced with a simple check against > the given flatness tolerance. > > Wine-Bug: https://bugs.winehq.org/show_bug.cgi?id=52492 > Signed-off-by: Robert Feuerbach rjfeuerbach@gmail.com > --- > dlls/gdiplus/graphicspath.c | 11 ++++++----- > 1 file changed, 6 insertions(+), 5 deletions(-) > > diff --git a/dlls/gdiplus/graphicspath.c b/dlls/gdiplus/graphicspath.c > index ce2666eedab..c1bdf226e52 100644 > --- a/dlls/gdiplus/graphicspath.c > +++ b/dlls/gdiplus/graphicspath.c > @@ -109,7 +109,7 @@ static INT path_list_count(path_list_node_t *node) > * - (x2, y2): first control point; > * - (x3, y3): second control point; > * - end : pointer to end point node > - * - flatness: admissible error of linear approximation. > + * - flatness: admissible error of linear approximation in coordinate units. > * > * Return value: > * TRUE : success > @@ -142,12 +142,13 @@ static BOOL flatten_bezier(path_list_node_t *start, REAL x2, REAL y2, REAL x3, R > mp[2].X = (mp[1].X + mp[3].X) / 2.0; > mp[2].Y = (mp[1].Y + mp[3].Y) / 2.0; > > - if ((x2 == mp[0].X && y2 == mp[0].Y && x3 == mp[1].X && y3 == mp[1].Y) || > - (x2 == mp[3].X && y2 == mp[3].Y && x3 == mp[4].X && y3 == mp[4].Y)) > - return TRUE; > - > pt = end->pt; > pt_st = start->pt; > + /* test for closely spaced points to avoid limited-precision errors in flatness check */ > + if((fabs(pt.X - mp[2].X) + fabs(pt.Y - mp[2].Y) + > + fabs(pt_st.X - mp[2].X) + fabs(pt_st.Y - mp[2].Y) ) <= flatness) > + return TRUE; > + > /* check flatness as a half of distance between middle point and a linearized path */ > if(fabs(((pt.Y - pt_st.Y)*mp[2].X + (pt_st.X - pt.X)*mp[2].Y + > (pt_st.Y*pt.X - pt_st.X*pt.Y))) <= > -- > 2.32.0 (Apple Git-132) > >
The flatness check is consistent with deriving from: dist_m = | dot(pt_st - mp[2], cross(pt_st, pt)) | / | pt_st - pt |
The simplified closeness check also just compares the same points on the curve: pt, pt_st, and mp[2].
It is possible, for example an S-shaped curve, for mp[2], pt, and pt_st to be collinear but for the true curve to vary more than the flatness. A more pathological case would be for all the control points [pt, pt_st, (x2,y2), (x3, y3)] to be colinear with (x2, y2) and (x3, y3) outside of the pt --> pt_st segment.
Taking these into account are deeper possible fixes than I was aiming for here, which was simply to improve the handling to prevent a crash in an application.
On Fri, Feb 4, 2022 at 2:12 PM Nikolay Sivov nsivov@codeweavers.com wrote:
On 2/4/22 20:33, Esme Povirk (she/they) wrote:
I'm not convinced by this approach. There might be a situation where the midpoint between the endpoints happens to be close to the midpoint between the control points, but the control points are further than the flatness, leaving a possibility that we stop while our approximation has more error than the flatness.
I feel like we should be testing the distance between the control points and the line segment (or the distance between the control points and an endpoint, which should be no less than the distance from the line segment).
I don't really understand the existing "half of distance between middle point and a linearized path" test either. Nikolay, can you explain this?
It's be a while. The idea is to check a distance from mp[2] point to {pt, pt_st} segment (dist_m), and stop recursion if it's small enough.
I haven't verified right now, but expression I believe was meant to get that distance as:
sqr(dot(mp - pt, pt_st - pt)) + sqr(dist_m) = sqr(dist(pt_st, pt));
On Fri, Feb 4, 2022 at 9:59 AM Robert Feuerbach rjfeuerbach@gmail.com
wrote:
The float equality and flatness calculation in flatten_bezier can fail due to the limited precision of the float math. The equality test was replaced with a simple check against the given flatness tolerance.
Wine-Bug: https://bugs.winehq.org/show_bug.cgi?id=52492 Signed-off-by: Robert Feuerbach rjfeuerbach@gmail.com
dlls/gdiplus/graphicspath.c | 11 ++++++----- 1 file changed, 6 insertions(+), 5 deletions(-)
diff --git a/dlls/gdiplus/graphicspath.c b/dlls/gdiplus/graphicspath.c index ce2666eedab..c1bdf226e52 100644 --- a/dlls/gdiplus/graphicspath.c +++ b/dlls/gdiplus/graphicspath.c @@ -109,7 +109,7 @@ static INT path_list_count(path_list_node_t *node)
- (x2, y2): first control point;
- (x3, y3): second control point;
- end : pointer to end point node
- flatness: admissible error of linear approximation.
- flatness: admissible error of linear approximation in coordinate
units.
- Return value:
- TRUE : success
@@ -142,12 +142,13 @@ static BOOL flatten_bezier(path_list_node_t
*start, REAL x2, REAL y2, REAL x3, R
mp[2].X = (mp[1].X + mp[3].X) / 2.0; mp[2].Y = (mp[1].Y + mp[3].Y) / 2.0;
- if ((x2 == mp[0].X && y2 == mp[0].Y && x3 == mp[1].X && y3 ==
mp[1].Y) ||
(x2 == mp[3].X && y2 == mp[3].Y && x3 == mp[4].X && y3 ==mp[4].Y))
return TRUE;pt = end->pt; pt_st = start->pt;
- /* test for closely spaced points to avoid limited-precision
errors in flatness check */
- if((fabs(pt.X - mp[2].X) + fabs(pt.Y - mp[2].Y) +
fabs(pt_st.X - mp[2].X) + fabs(pt_st.Y - mp[2].Y) ) <=flatness)
return TRUE;/* check flatness as a half of distance between middle point anda linearized path */
if(fabs(((pt.Y - pt_st.Y)*mp[2].X + (pt_st.X - pt.X)*mp[2].Y + (pt_st.Y*pt.X - pt_st.X*pt.Y))) <=-- 2.32.0 (Apple Git-132)
My apologies, the equation of the flatness check is consistent with: dist_m = | cross(pt_st - pt, mp[2] - pt) | / | pt_st - pt | < 0.5 * flatness
On Fri, Feb 4, 2022 at 2:50 PM Robert Feuerbach rjfeuerbach@gmail.com wrote:
The flatness check is consistent with deriving from: dist_m = | dot(pt_st - mp[2], cross(pt_st, pt)) | / | pt_st - pt |
The simplified closeness check also just compares the same points on the curve: pt, pt_st, and mp[2].
It is possible, for example an S-shaped curve, for mp[2], pt, and pt_st to be collinear but for the true curve to vary more than the flatness. A more pathological case would be for all the control points [pt, pt_st, (x2,y2), (x3, y3)] to be colinear with (x2, y2) and (x3, y3) outside of the pt --> pt_st segment.
Taking these into account are deeper possible fixes than I was aiming for here, which was simply to improve the handling to prevent a crash in an application.
On Fri, Feb 4, 2022 at 2:12 PM Nikolay Sivov nsivov@codeweavers.com wrote:
On 2/4/22 20:33, Esme Povirk (she/they) wrote:
I'm not convinced by this approach. There might be a situation where the midpoint between the endpoints happens to be close to the midpoint between the control points, but the control points are further than the flatness, leaving a possibility that we stop while our approximation has more error than the flatness.
I feel like we should be testing the distance between the control points and the line segment (or the distance between the control points and an endpoint, which should be no less than the distance from the line segment).
I don't really understand the existing "half of distance between middle point and a linearized path" test either. Nikolay, can you explain this?
It's be a while. The idea is to check a distance from mp[2] point to {pt, pt_st} segment (dist_m), and stop recursion if it's small enough.
I haven't verified right now, but expression I believe was meant to get that distance as:
sqr(dot(mp - pt, pt_st - pt)) + sqr(dist_m) = sqr(dist(pt_st, pt));
On Fri, Feb 4, 2022 at 9:59 AM Robert Feuerbach rjfeuerbach@gmail.com
wrote:
The float equality and flatness calculation in flatten_bezier can fail due to the limited precision of the float math. The equality test was replaced with a simple check against the given flatness tolerance.
Wine-Bug: https://bugs.winehq.org/show_bug.cgi?id=52492 Signed-off-by: Robert Feuerbach rjfeuerbach@gmail.com
dlls/gdiplus/graphicspath.c | 11 ++++++----- 1 file changed, 6 insertions(+), 5 deletions(-)
diff --git a/dlls/gdiplus/graphicspath.c b/dlls/gdiplus/graphicspath.c index ce2666eedab..c1bdf226e52 100644 --- a/dlls/gdiplus/graphicspath.c +++ b/dlls/gdiplus/graphicspath.c @@ -109,7 +109,7 @@ static INT path_list_count(path_list_node_t *node)
- (x2, y2): first control point;
- (x3, y3): second control point;
- end : pointer to end point node
- flatness: admissible error of linear approximation.
- flatness: admissible error of linear approximation in
coordinate units.
- Return value:
- TRUE : success
@@ -142,12 +142,13 @@ static BOOL flatten_bezier(path_list_node_t
*start, REAL x2, REAL y2, REAL x3, R
mp[2].X = (mp[1].X + mp[3].X) / 2.0; mp[2].Y = (mp[1].Y + mp[3].Y) / 2.0;
- if ((x2 == mp[0].X && y2 == mp[0].Y && x3 == mp[1].X && y3 ==
mp[1].Y) ||
(x2 == mp[3].X && y2 == mp[3].Y && x3 == mp[4].X && y3 ==mp[4].Y))
return TRUE;pt = end->pt; pt_st = start->pt;
- /* test for closely spaced points to avoid limited-precision
errors in flatness check */
- if((fabs(pt.X - mp[2].X) + fabs(pt.Y - mp[2].Y) +
fabs(pt_st.X - mp[2].X) + fabs(pt_st.Y - mp[2].Y) ) <=flatness)
return TRUE;/* check flatness as a half of distance between middle point anda linearized path */
if(fabs(((pt.Y - pt_st.Y)*mp[2].X + (pt_st.X - pt.X)*mp[2].Y + (pt_st.Y*pt.X - pt_st.X*pt.Y))) <=-- 2.32.0 (Apple Git-132)
Hi,
While running your changed tests, I think I found new failures. Being a bot and all I'm not very good at pattern recognition, so I might be wrong, but could you please double-check?
Full results can be found at: https://testbot.winehq.org/JobDetails.pl?Key=107024
Your paranoid android.
=== debian11 (64 bit WoW report) ===
Report validation errors: gdiplus:graphicspath has no test summary line (early exit of the main process?) gdiplus:graphicspath has unaccounted for todo messages
Tests should pass (except for wine_todo's) at every point in the series. Given that the problem you're testing is a stack overflow, I don't think there's any way you can test it until it's fixed, so the test will need to be sent after the fix, or in the same patch.
Thank you for the clarification. I'll send in the patch for the test after the discussion about the fix is done. It is interesting this stack overflow only happens on one of the build platforms (64-bit debian), and of course on the 64-bit MacOS 11.6.3 I'm using.
On Fri, Feb 4, 2022 at 12:37 PM Esme Povirk (she/they) esme@codeweavers.com wrote:
Tests should pass (except for wine_todo's) at every point in the series. Given that the problem you're testing is a stack overflow, I don't think there's any way you can test it until it's fixed, so the test will need to be sent after the fix, or in the same patch.
-
Esme Povirk (she/they) -
Fabian Maurer -
Marvin -
Nikolay Sivov -
Robert Feuerbach