[Date Prev][Date Next][Thread Prev][Thread Next][Author Index][Date Index][Thread Index]

RE>cfront bug/new binary



Reply to:   RE>cfront bug/new binary
}People with stand-alone sun4s take note:
}
}Today I found and patched a bug in cfront that imposed a n<128 limit on the
}number of virtual functions in a class and all of its super classes.  As near
}as I understand the relevant piece of cfront, we should be ok upto about 228
}functions.  I put the repaired version in /files5/nat2.0/sun4/cfront2, and
}renamed the old one as cfront2.vtblbug.  The present version of the binary
}allows the core file from cfront crashes to appear, so if anyone manages to
}crash it, could you please point me at it.  The new version successfully
}compiled all the files that crashed it before--but who knows what evil lurks?
}Also, I compiled it with heavier optimization than previously, so people
}should notice slightly faster compiles.

Great work Eric!

except...
  seems the pure virtual bug is back.  The stuff I was compiling on Rogers
machine had a bug which appears to be the pure virtual bug. (attempted to
access pure virtual function from a constructor)
Other than that, all's great!

--Hugh