On Thu, 2013-04-25 at 14:23 -0300, Sylvain Fournier wrote: > Andrew, > > Thanks for your quick answer. I will wait for the patch. > >
Looks like the bug is in the original Fortran code. I wrote your instance after preprocessing by glp_relax4 to a text file and run the original RELAX-IV code: 0001 // job relax4 0002 // option subchk 0003 // exec fortran 3947 // exec fortran 3952 // exec link 3953 // assgn RELAX4.INP relax.txt 3954 // exec * READ PROBLEM DATA FROM RELAX4.INP END OF READING NUMBER OF NODES = 6328, NUMBER OF ARCS = 167591 CONSTRUCT LINKED LISTS FOR THE PROBLEM ENTER THE INITIALIZATION DESIRED <0> FOR THE DEFAULT INITIALIZATION <1> FOR AUCTION INITIALIZATION CALLING RELAX4 TO SOLVE THE PROBLEM *********************************** Subscript out of range on file line 3186, procedure ascnt1. Attempt to access the 10001-th element of variable prdcsr. SIGABRT signal raised, execution terminated Call traceback 004302A9 00000011 00430298 stdc.lib(xabort.o) 0043323D 000000B1 0043318C stdc.lib(raise.o) 0042FB27 0000003B 0042FAEC stdc.lib(abort.o) 0042C3E5 00000051 0042C394 f2c.lib(sig_die.o) 0042840F 000000B7 00428358 f2c.lib(s_rnge.o) 0041CC65 0001BB81 004010E4 file0003.o 0040FABE 0000E9DA 004010E4 file0003.o 00401D06 00000C22 004010E4 file0003.o 00426440 00000054 004263EC f2c.lib(main.o) 004010C1 000000C1 00401000 crt0.o End of traceback Program abnormally terminated RELAX4.rps:3954: program exit code is 255 (0xFF) This is exactly the same array which is clobbered and exactly the same place where I detected the bug in the C code. I need a time to check everything more carefully. Andrew Makhorin _______________________________________________ Bug-glpk mailing list [email protected] https://lists.gnu.org/mailman/listinfo/bug-glpk
