• sito (11/11) Jul 17 2005 DMD 0.128, linux.
• Adrian Ratnapala (9/20) Jul 27 2005 This is right. The new version of exists() for
• Nick (11/15) Aug 06 2005 Right. Here is a short patch on this one too:

sito <sito htf.highway.ne.jp> writes:

DMD 0.128, linux.
-- exists_bug.d --
import std.file;

void main()
{
assert(exists("."));
}
--

$ dmd exists_bug.d
$ ./exists_bug
Error: AssertError Failure exists_bug(5)

Adrian Ratnapala <adrianr physics.uq.edu.au> writes:

This is right.  The new version of exists() for
Linux uses the access() libc call, but interprets
the result the wrong way, it does:


return access(toStringz(name),0) != 0;

but access() returns zero on _success_,
-1 on failure.  Just removing the "!= 0" or
changing it to "!= -1" or "== 0" should
do the trick.

In article <dbe7c9$rsh$1 digitaldaemon.com>, sito says...
DMD 0.128, linux.
-- exists_bug.d --
import std.file;

void main()
{
assert(exists("."));
}
--

$ dmd exists_bug.d
$ ./exists_bug
Error: AssertError Failure exists_bug(5)

Nick <Nick_member pathlink.com> writes:

In article <dc7o4v$2g2a$1 digitaldaemon.com>, Adrian Ratnapala says...
This is right.  The new version of exists() for
Linux uses the access() libc call, but interprets
the result the wrong way, it does:
[snip]

Right. Here is a short patch on this one too:

--- file.d      2005-08-06 21:57:06.000000000 +0200
+++ filefixed.d 2005-08-07 08:54:31.133089424 +0200
   -804,7 +804,7   

int exists(char[] name)
{
-    return access(toStringz(name),0) != 0;
+    return access(toStringz(name),0) == 0;

/+
struct_stat statbuf;