Debugging the Code
Looking at the generated code
Using the FIR backend
The FIR (Faust Imperative Representation) backend can possibly be used to look at a textual version of the intermediate imperative language.
import("stdfaust.lib");
vol = hslider("volume [unit:dB]", 0, -96, 0, 0.1) : ba.db2linear : si.smoo;
freq1 = hslider("freq1 [unit:Hz]", 1000, 20, 3000, 1);
freq2 = hslider("freq2 [unit:Hz]", 200, 20, 3000, 1);
process = vgroup("Oscillator", os.osc(freq1) * vol, os.osc(freq2) * vol);
For instance compiling the previous code with the faust -lang fir osc.dsp command will display various statistics, for example the number of operations done in the generated compute method:
======= Compute DSP begin ==========
Instructions complexity : Load = 23 Store = 9
Binop = 12 [ { Int(+) = 1 } { Int(<) = 1 } { Real(*) = 3 } { Real(+) = 5 } { Real(-) = 2 } ]
Mathop = 2 [ { floorf = 2 } ]
Numbers = 8
Declare = 1
Cast = 2
Select = 0
Loop = 1
As well as the DSP structure memory size and layout, and read/write statistics:
======= Object memory footprint ==========
Heap size int = 4 bytes
Heap size int* = 0 bytes
Heap size real = 48 bytes
Total heap size = 68 bytes
Stack size in compute = 28 bytes
======= Variable access in compute control ==========
Field = fSampleRate size = 1 r_count = 0 w_count = 0
Field = fConst1 size = 1 r_count = 1 w_count = 0
Field = fHslider0 size = 1 r_count = 1 w_count = 0
Field = fConst2 size = 1 r_count = 0 w_count = 0
Field = fRec0 size = 2 r_count = 0 w_count = 0
Field = fConst3 size = 1 r_count = 2 w_count = 0
Field = fHslider1 size = 1 r_count = 1 w_count = 0
Field = fRec2 size = 2 r_count = 0 w_count = 0
Field = fHslider2 size = 1 r_count = 1 w_count = 0
Field = fRec3 size = 2 r_count = 0 w_count = 0
======= Variable access in compute DSP ==========
Field = fSampleRate size = 1 r_count = 0 w_count = 0
Field = fConst1 size = 1 r_count = 0 w_count = 0
Field = fHslider0 size = 1 r_count = 0 w_count = 0
Field = fConst2 size = 1 r_count = 1 w_count = 0
Field = fRec0 size = 2 r_count = 4 w_count = 2
Field = fConst3 size = 1 r_count = 0 w_count = 0
Field = fHslider1 size = 1 r_count = 0 w_count = 0
Field = fRec2 size = 2 r_count = 4 w_count = 2
Field = fHslider2 size = 1 r_count = 0 w_count = 0
Field = fRec3 size = 2 r_count = 4 w_count = 2
Those informations can possibly be used to detect abnormal memory consumption.
Debugging the DSP Code
On a computer, doing a computation that is undefined in mathematics (like val/0 of log(-1)), and unrepresentable in floating-point arithmetic, will produce a NaN value, which has a special internal representation. Similarly, some computations will exceed the range that is representable with floating-point arithmetics, and are represented with a special INFINITY value, which value depends of the choosen type (like float, double or long double).
After being produced, those values can actually contaminate the following flow of computations (that is Nan + any value = NaN for instance) up to the point of producing incorrect indexes when used in array access, and causing memory access crashes.
The Faust compiler gives error messages when the written code is not syntactically or semantically correct, and the interval computation system on signals is supposed to detect possible problematic computations at compile time, and refuse to compile the corresponding DSP code. But the interval calculation is currently quite imperfect, can misbehave, and possibly allow problematic code to be generated.
Several strategies have been developed to help programmers better understand their written DSP code, and possibly analyse it, both at compile time and runtime.
Debugging at compile time
The -ct option
Using the -ct compilation option allows to check table index range and generate safe table access code. It verifies that the signal range is compatible with the table size, and if needed, generate safe read and write indexes access code, by constraining them to stay in a given [0.. size-1] range.
Note that since the signal interval calculation is imperfect, you may see false positives, and uneeded range constraining code might be generated, especially when using recursive signals where the interval calculation system will typically produce [-inf, inf] range, which is not precise enough to correctly describe the real signal range.
The -me option
Starting with version 2.37.0, mathematical functions which have a finite domain (like sqrt defined for positive or null values, or asin defined for values in the [-1..1] range) are checked at compile time when they actually compute values at that time, and raise an error if the program tries to compute an out-of-domain value. If those functions appear in the generated code, their domain of use can also be checked (using the interval computation system) and the -me option will display warnings if the domain of use is incorrect. Note that again because of the imperfect interval computation system, false positives may appear and should be checked.
Debugging at runtime
The interp-tracer tool
The interp-tracer tool runs and instruments the compiled program using the Interpreter backend. Various statistics on the code are collected and displayed while running and/or when closing the application, typically FP_SUBNORMAL, FP_INFINITE and FP_NAN values, or INTEGER_OVERFLOW, CAST_INT_OVERFLOW and DIV_BY_ZERO operations, or LOAD/STORE errors.
See the complete documentation and the Advanced debugging with interp-tracer tutorial.
The faust2caqt tool
On macOS, the faust2caqt script has a -me option to catch math computation exceptions (floating point exceptions and integer div-by-zero or overflow, etc.) at runtime. Developers can possibly use the dsp_me_checker class to decorate a given DSP object with the math computation exception handling code.
Fixing the errors
These errors must then be corrected by carefully checking signal range, like verifying the min/max values in vslider/hslider/nentry user-interface items.
Additional Resources
Note that the Faust math library contains the implementation of isnan and isinf functions that may help during development.
Handling infinity and not-a-number (NaN) the right way still remains a tricky problem that is not completely handled in the current version of the compiler. Dario Sanfilippo blog post is a very helpful summary of the situation with a lot of practical solutions to write safer DSP code.