There are quite a few regressions (336 test cases, 2774 configurations) in DMD-0.153, however most - if not all - seem to have a common cause:

(sample source: http://dstress.kuehne.cn/run/t/typeid_86_C.d)


compiling without "-g"
=====================
Program received signal SIGSEGV, Segmentation fault.
0x0805030a in _init__Dmain1B ()
(gdb) bt
#0  0x0805030a in _init__Dmain1B ()
#1  0x00000012 in ?? ()
#2  0xffffbd48 in ?? ()
#3  0x08049a19 in _Dmain ()
Previous frame inner to this frame (corrupt stack?)


compiling with "-g"
===================
(gdb) break _init__Dmain1B
Breakpoint 1 at 0x80502cd
(gdb) run
Starting program: ./typeid_86_C

Program exited normally.


compiling without "-g" and manual linking without "--gc-sections"
=================================================================
Starting program: ./typeid_86_C
Error: AssertError Failure run/t/typeid_86_C.d(18)

Program exited with code 01.


Thomas

Seems to be the same problem as bugzilla entry 100: assert error. Exception handling is not working under linux when linked with that --gc-sections flag?

L.

Lionello Lunesu wrote:
> Seems to be the same problem as bugzilla entry 100: assert error. Exception handling is not working under linux when linked with that --gc-sections flag?
> 
> L.

Same w/ bug #101. The --gc-sections flag is breaking pre and post contracts.