From: ts <decoux@...>
Date: 2004-05-25T01:15:56+09:00
Subject: Re: compiling mmap?

>>>>> "M" == Mark Hubbart <discord@mac.com> writes:

M> test_13_div(TestMmap) [mmapt.rb:304]:
M> empty.
M> <true> expected but was
M> <false>.

 Well, it's normal : the test is not adapted with "rw" it will not truncate
 the file. This is not the good correction, try with

svg% diff -u mmap.c~ mmap.c
--- mmap.c~     2004-05-24 11:20:04.000000000 +0200
+++ mmap.c      2004-05-24 18:13:45.000000000 +0200
@@ -420,10 +420,7 @@
        }
        else if (strcmp(mode, "w") == 0) {
            smode = O_RDWR | O_TRUNC;
-           pmode = PROT_WRITE;
-           if (!NIL_P(fdv)) {
-               pmode |= PROT_READ;
-           }
+           pmode = PROT_READ | PROT_WRITE;
        }
        else if (strcmp(mode, "rw") == 0 || strcmp(mode, "wr") == 0) {
            smode = O_RDWR;
svg% 

 and in tests/mmapt.rb reverse the modification

   assert_kind_of(Mmap, m1 = Mmap.new("tmp/bb", "w"), "new a")

 Now, why it need PROT_READ ???



Guy Decoux