ID0001081: **This issue was created automatically from Mantis Issue 1081. Further discussion may take place here.** --- | **Id** | **Project** | **Category** | **View** | **Due Date** | **Updated** | | --- | --- | --- | --- | --- | --- | | ID0001081 | Frama-C | Kernel | public | 2012-02-05 | 2012-09-19 | | | | | | | | | --- | --- | --- | --- | --- | --- | | **Reporter** | pascal | **Assigned To** | virgile | **Resolution** | fixed | | **Priority** | normal | **Severity** | minor | **Reproducibility** | always | | **Platform** | - | **OS** | - | **OS Version** | - | | **Product Version** | Frama-C Nitrogen-20111001 | **Target Version** | - | **Fixed in Version** | Frama-C Oxygen-20120901 | ### Description : The program below should probably be rejected as ill-typed: main(){ int foo, x, y; foo ? (void)x : (signed char)y; return 0; } but: $ ppc/bin/toplevel.opt -val t.c [kernel] preprocessing with "gcc -C -E -I. t.c" ... [value] Values at end of function main: __retres ? {0} Incidentally, for this example, the access to uninitialized variable foo should also cause the program to be analyzed as non-terminating. Not sure whether that secondary bug is in the front-end or in the value analysis.