Shooting yourself in the foot, vs blowing your whole damned leg off.
-
I don't like using assert() because it simply causes a bootloop on these devices when hit. I don't have a debugger. Also this code did not originate with me. I forked it from another project, and have done a lot of work to it, but a significant portion I have yet to touch.
Check out my IoT graphics library here: https://honeythecodewitch.com/gfx And my IoT UI/User Experience library here: https://honeythecodewitch.com/uix
May I inquire means by which garbage value originated.
-
May I inquire means by which garbage value originated.
solar flares probably. It's what happens in C and C++ when you don't initialize a value explicitly on a release build. It's garbage. leftover bits in stale ram.
Check out my IoT graphics library here: https://honeythecodewitch.com/gfx And my IoT UI/User Experience library here: https://honeythecodewitch.com/uix
-
solar flares probably. It's what happens in C and C++ when you don't initialize a value explicitly on a release build. It's garbage. leftover bits in stale ram.
Check out my IoT graphics library here: https://honeythecodewitch.com/gfx And my IoT UI/User Experience library here: https://honeythecodewitch.com/uix
Don't static analyzers catch that sort of thing.
-
Don't static analyzers catch that sort of thing.
It's not easy to run static analysis on this code because I don't have much control over my build environment. It's all controlled and automagic via PlatformIO. That's often nice because it streamlines my workflow but limits the sorts of tools i can use.
Check out my IoT graphics library here: https://honeythecodewitch.com/gfx And my IoT UI/User Experience library here: https://honeythecodewitch.com/uix
-
It's not easy to run static analysis on this code because I don't have much control over my build environment. It's all controlled and automagic via PlatformIO. That's often nice because it streamlines my workflow but limits the sorts of tools i can use.
Check out my IoT graphics library here: https://honeythecodewitch.com/gfx And my IoT UI/User Experience library here: https://honeythecodewitch.com/uix
Hmm. Just a suggestion, but I've used static analyzers such as those included in VS2022 against code from embedded environments more than once. True, it can be a PITA to set up and you have to separate the wheat from the chaff. The end result however can catch a lot of simple errors that aren't detectable by less comprehensive compilers.
Software Zen:
delete this; -
Edit: Actual photographic evidence of victory non-trivial SVG document[^]
parse_res.attr[0].strokeDashCount = 0;
It has taken me a day of extremely stressful hair pulling to find that this line of code didn't exist and it should have. It was manifesting as a stack overflow, so I spent my day, tracking that, and pruning the stack. Here was the offending code that caused the memory corruption that reported as a stack issue.
for (i = 0; i < attr->strokeDashCount; i++)
shape->stroke_dash_array[i] = attr->strokeDashArray[i] * scale;Here the strokeDashCount value was a ridiculously high garbage number, with (un)predictable results. This never would have happened in a high level language. I love me some C and C++. It's an incredible programming language family, but also a very powerful weapon you can wield against yourself if you're not careful.
Check out my IoT graphics library here: https://honeythecodewitch.com/gfx And my IoT UI/User Experience library here: https://honeythecodewitch.com/uix
-
honey the codewitch wrote:
ridiculously high
Are you sayin' uninitialized?
"If we don't change direction, we'll end up where we're going"
yes, which in this case resulted in a very high value. (The machine is realtime, which means the value is the same every time it boots, even though it's "garbage")
Check out my IoT graphics library here: https://honeythecodewitch.com/gfx And my IoT UI/User Experience library here: https://honeythecodewitch.com/uix
-
Hmm. Just a suggestion, but I've used static analyzers such as those included in VS2022 against code from embedded environments more than once. True, it can be a PITA to set up and you have to separate the wheat from the chaff. The end result however can catch a lot of simple errors that aren't detectable by less comprehensive compilers.
Software Zen:
delete this;It's far easier to keep my lib cross platform, run static analysis on what I can on the PC, and furble with the rest on embedded blindly if needed than it is to try to make PIO's toolchain work outside the PIO IDE.
Check out my IoT graphics library here: https://honeythecodewitch.com/gfx And my IoT UI/User Experience library here: https://honeythecodewitch.com/uix
-
yes, which in this case resulted in a very high value. (The machine is realtime, which means the value is the same every time it boots, even though it's "garbage")
Check out my IoT graphics library here: https://honeythecodewitch.com/gfx And my IoT UI/User Experience library here: https://honeythecodewitch.com/uix
-
Which baffled me, since it that kind of mistake sounds soo different from your usual meticulous self !
"If we don't change direction, we'll end up where we're going"
This particular code is really complicated. Look at this mess
typedef struct NSVGattrib {
char id[64];
float xform[6];
gfx::rgba_pixel<32> fillColor;
gfx::rgba_pixel<32> strokeColor;
float opacity;
float fillOpacity;
float strokeOpacity;
char fillGradient[64];
char strokeGradient[64];
float strokeWidth;
float strokeDashOffset;
float strokeDashArray[NSVG_MAX_DASHES];
int strokeDashCount;
svg_line_join strokeLineJoin;
svg_line_cap strokeLineCap;
float miterLimit;
svg_fill_rule fillRule;
float fontSize;
gfx::rgba_pixel<32> stopColor;
float stopOpacity;
float stopOffset;
char hasFill;
char hasStroke;
char visible;
} NSVGattrib;
using reader_t = ml_reader_ex<2048>;
struct svg_css_class {
char selector[512];
char* value;
svg_css_class* next;
};
struct svg_parse_result {
reader_t* reader;
svg_css_class* css_classes;
svg_css_class* css_class_tail;
void* (*allocator)(size_t);
void* (*reallocator)(void*, size_t);
void (*deallocator)(void*);
char lname[32];
char aname[512];
char avalue[512];
char style_val[256];
char class_val[128];
char* d;
size_t d_size;
NSVGattrib attr[NSVG_MAX_ATTR];
int attrHead;
float* pts;
int npts;
int cpts;
svg_path* plist;
size_t image_size;
svg_image_info* image;
NSVGgradientData* gradients;
svg_shape* shapesTail;
float viewMinx, viewMiny, viewWidth, viewHeight;
int alignX, alignY, alignType;
float dpi;
char pathFlag;
char defsFlag;
};I highlighted the portion of the inner structure I forgot to initialize (the head attribute, as the rest are copied from that)
Check out my IoT graphics library here: https://honeythecodewitch.com/gfx And my IoT UI/User Experience library here: https://honeythecodewitch.com/uix
