Unknown crash with DifEq

[César Pérez Álvarez]

When we compile this code:

---

Set RandNoise(Date ini, Date end, TimeSet dating)
{

Serie residuals = SubSer(Gaussian(0, 0.10, dating), ini, end);
Polyn ma = (1-0.6*B)*(1-0.5*B^{7);
Polyn ar = (1-0.4*B);
Polyn dif = (1-B)*(1-B}7)*(1-B^365);

Date iniRes = Succ(ini, dating, -Degree(ma));
Date endRes = Succ(ini, dating, -1);
Serie iniResiduals = SubSer(Gaussian(0, 0.10, dating), iniRes, endRes);
Serie noise = DifEq(ma/(dif*ar), iniResiduals<<residuals,8);
SetOfSerie(ini, residuals)

};

Set test = RandNoise(y2000m01d01, y2007m01d01, Diario);

---

TOL crashes. Its obvious that we have an eror in "SetOfSerie(ini, residuals)" because "ini" is a "Date" when it must be a "Serie" but TOL never must crash.

If we try with this another code:

---

Set RandNoise(Date ini, Date end, TimeSet dating)
{

Serie residuals = SubSer(Gaussian(0, 0.10, dating), ini, end);
Polyn ma = (1-0.6*B)*(1-0.5*B^{7);
Polyn ar = (1-0.4*B);
Polyn dif = (1-B)*(1-B}7)*(1-B^365);

Date iniRes = Succ(ini, dating, -Degree(ma));
Date endRes = Succ(ini, dating, -1);
Serie iniResiduals = SubSer(Gaussian(0, 0.10, dating), iniRes, endRes);

Serie noise = DifEq(ma/(dif*ar), iniResiduals<<residuals,8);

SetOfSerie(ini, residuals)

};

Set test = RandNoise(y2000m01d01, y2007m01d01, Diario);

---

TOL does not crash, and a correct warning is advise us of our error.

I can not isolate much more the error, sorry.

[Víctor de Buen Remiro]

We are working about this bug
Thanks to report it

[Víctor de Buen Remiro]

Bug has been fixed in CVS (HEAD and 1.1.6 branches)

BTsrPolyn BTsrDifEq::ser_ must be a pointer to avoid default copy operator that jumps PutDating and doesn't increments reference of dating_

Congratulations!
This is a very old problem and this bug report has been a great help to detect it.

A standard test has been added at ​http://cvs.tol-project.org/viewcvs.cgi/tol_tests/tol/Bugzilla/bug_000535
Thanks