Still working on my project of a D wrapped for libmpdec
(https://www.bytereef.org/mpdecimal/).
Consider the following code:
void main()
{
mpd_context_t ctx;
mpd_t* a;
mpd_ieee_context(&ctx, 128);
a= mpd_new(&ctx);
}
It seems to work fine. mpd_new is a C function for allocating a
new mpdecimal object via
malloc, prototype:
mpd_t *mpd_new(mpd_context_t *ctx);
see the documentation at:
https://www.bytereef.org/mpdecimal/doc/libmpdec/decimals.html?highlight=mpd_new#std:topic-mpd_new
However, if I put the very same code in the unittest section of a
module, it fails with
a segmentation fault and I don't understand why!
Personally, I found the unittest mechanism rather obscure.
For instance I have tried to debug the problem using gdb and I
found that the following code was run
module dub_test_root;
2 import std.typetuple;
3 static import mpdec.decimal;
4 static import mpdec.mpdec;
5 alias allModules = TypeTuple!(mpdec.decimal, mpdec.mpdec);
6
7 import std.stdio;
8 import core.runtime;
9
10 void main() { writeln("All unit tests have been run
successfully."); }
11 shared static this() {
12 version
(Have_tested) {
13 import
tested;
14 import
core.runtime;
15 import
std.exception;
16
Runtime.moduleUnitTester = () => true;
17 enforce(runUnitTests!allModules(new
ConsoleTestResultWriter), "Unit tests failed.");
18 }
19 }
20
I didn't even new that this code was there! (and I didn't manage
to set a breakpoint
in my unit test, is there a way to do that?). I rather prefer a
separate test program!
