Segment7: Update on "undefined method for Fixnum"

Update on "undefined method for Fixnum"

drbrain@segment7.net (Eric Hodel) — Thu, 20 Apr 2006 11:37:44 -0700

A week or so ago I managed to chase down the “undefined method for Fixnum” bug to the compiler flags -g -O1 -pipe -fno-common -DRUBY_EXPORT -fschedule-insns2. For now you can get rid of the bug by compiling -O1 instead of switching to gcc3. I haven’t had time to track it down further, mostly because I ran out of time and the man page doesn’t specify what -O1 really contains.

Hopefully some smarter eyeballs than mine can jump on this and figure out if GCC is wrong or if Ruby is missing a volatile somewhere.

"Update on "undefined method for Fixnum"" by Daniel Berger

Fri, 21 Apr 2006 06:39:13 -0700

zenspider, sorry, I meant the problem with -fschedule-insns2 in particular. I could have sworn this error has cropped up on other platforms in the past. Maybe those were compiler bugs as well. :/

"Update on "undefined method for Fixnum"" by zenspider

Fri, 21 Apr 2006 01:01:04 -0700

Dan, yes to both gcc and architecture. We’ve been chasing this for a while now.

"Update on "undefined method for Fixnum"" by Daniel Berger

Thu, 20 Apr 2006 13:43:12 -0700

From what I can gather from Google, this appears to be a bug in gcc 4.x on OS X. What version are you using? And what architecture is your machine?

"Update on "undefined method for Fixnum"" by Caleb

Thu, 20 Apr 2006 13:14:28 -0700

way over my head.

but thanks!

"Update on "undefined method for Fixnum"" by Daniel Berger

Thu, 20 Apr 2006 12:54:08 -0700

According to my (slightly outdated) GCC book, -O1 means it compiles with the following flags:

-fno_optimize_size
-fdefer_pop
-fthread_jumps
-jguess_branch_prob
-cprop-registers
-fdelayed_branch
-fomit_frame_pointer

The -fomit_frame_pointer flag is set only if the debugger is able to work without it on the current platform.

Hope that helps,

Dan