Sender: rich AT phekda DOT freeserve DOT co DOT uk
Message-ID: <3E2943DA.523770FC@phekda.freeserve.co.uk>
Date: Sat, 18 Jan 2003 12:08:58 +0000
From: Richard Dawe <rich AT phekda DOT freeserve DOT co DOT uk>
X-Mailer: Mozilla 4.77 [en] (X11; U; Linux 2.2.23 i586)
X-Accept-Language: de,fr
MIME-Version: 1.0
To: djgpp-workers AT delorie DOT com
Subject: Re: Take on __solve_symlinks()
References: <3E286C07 DOT 7040200 AT mif DOT vu DOT lt>
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
Reply-To: djgpp-workers AT delorie DOT com

Hello.

Laurynas Biveinis wrote:
> Well, maybe I'm on drugs,

Maybe the problem is that I'm not on drugs (caffeine)! ;)

> but I was unable to find good testcases for debugging it. It passes with
> flying colours every single test I made, except for this case:
> 
>     symlink("c:/autoexec.bat", "/link");
>     test_success("/link", "c:/autoexec.bat");
> 
> I.e. it didn't resolve /link properly. But I failed to create any more
> failing tests, including situations described by you. Maybe you could
> give me the testcase a patch to tests/libc/compat/unistd/xsymlink.c or
> precisely describe particular directory and file layout for uncovering
> the bugs?

I reverted my changes and I'm now running fileutils, sh-utils and textutils
built with DJGPP CVS HEAD. But the test program still fails for me:
 
bash-2.04$ pwd
c:/djgpp.204/tests/libc/compat/unistd
bash-2.04$ ./xsymlink.exe
Running __solve_symlinks() and readlink() testsuite:
Tests that check __solve_symlinks() works:
Test 1 passed
Test 2 passed
Test 3 passed
Test 4 passed
Test 5 passed
Test 6 passed
Test 7 passed
Test 8 passed
Test 9 passed
Test 10 passed
Test 11 passed
Test 12 passed
Test 13 passed
Tests that check __solve_symlinks() based on current directory:
Test 1: Solving
c:/djgpp.204/tests/libc/compat/unistd/../../../../../djgpp.204/t
ests/libc/compat/unistd/test1
Test 1 passed
Test 2: Solving c:../../../../../djgpp.204/tests/libc/compat/unistd/test1
Test 2 passed
Test 3: Solving c:../../../../../../djgpp.204/tests/libc/compat/unistd/test1
Test 3 failed - __solve_symlinks returns wrong resolved path
Returned path: c:/djgpp.204/tests/libc/compat/unistd/test1
Expected path: c:/djgpp.204/tests/libc/compat/unistd/file1
Tests that check __solve_symlinks() failure cases:
Test 1 passed
Test 2 passed
FAIL

Do you expect this test to fail?

Here some ideas for more tests:

* symlinks between drives;
* paths using /dev/<drive> instead of <drive>:;
* paths using environment variables in paths - e.g.: /dev/env/FOO.

Basically tests that mix absolute and relative paths in every possible way.
Most of the tests are too "normal" at the moment IMHO. Not that I am
criticising. It's just that you may not have thought of bizarre filenames as
much as I have, while testing fileutils, etc. ;)

> Below are the new testcases from testsuite I've used, and the patch I'm
> commiting in a second to fix /link case. The fix is trivial.
> 
>     symlink("c:/autoexec.bat", "/link");
>     test_success("c:/link", "c:/autoexec.bat");
>     test_success("/link", "c:/autoexec.bat");
>     test_success("c://link", "c:/autoexec.bat");
>     test_success("//link", "c:/autoexec.bat");
>     test_success("c:/./link", "c:/autoexec.bat");
>     test_success("/../link", "c:/autoexec.bat");
>     test_success("c:/../link", "c:/autoexec.bat");
>     test_success("/../../link", "c:/autoexec.bat");
>     test_success("c:/../../link", "c:/autoexec.bat");
>        /* FIXME below, but note that you're insane to run this test
> really that deep :) */
>     test_success("c:../../../../../../../../../../../../../../../link",
> "c:/autoexec.bat");
[snip]

Why is this marked as a FIXME?

Why is it insane? Sure, having more than 8 levels of directories is
unsupported on DOS, but the code should cope with excess ".."s.

The patch you posted looked good.

Thanks, bye, Rich =]

-- 
Richard Dawe [ http://www.phekda.freeserve.co.uk/richdawe/ ]