[Top][All Lists]
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: coreutils-5.96 bug report
From: |
Paul Eggert |
Subject: |
Re: coreutils-5.96 bug report |
Date: |
Sun, 11 Jun 2006 20:57:54 -0700 |
User-agent: |
Gnus/5.1008 (Gnus v5.10.8) Emacs/21.4 (gnu/linux) |
I tracked the problem down to the following test in
tests/readlink/can-e:
v=`readlink -e ${p}regfile` || break
test "$v" = "$my_pwd/$tmp/regfile" || break
Given the transcript you sent me privately, that last command turned
out to be equivalent to the following:
test
/Users/robingene/desktop/coreutils-5.96/tests/readlink/can-e.tmp/17775/regfile \
=
/Users/robingene/Desktop/coreutils-5.96/tests/readlink/can-e.tmp/17775/regfile
The two file names differ: one says "desktop" while the other says
"Desktop". There are similar problems in can-f and can-m.
Since you are using Darwin, I surmise that you're running the test on
an HFS or HFS+ file system, where accessing files by name is a case
insensitive process. The obvious workaround would be to run "make
check" in a UFS file system, since UFS is case sensitive.
The first file name came from "readlink -e", the second from "pwd".
I don't know why the two names would disagree on an HFSish file
system; perhaps someone with some old MacOS expertise (not me!) can
chime in.
Thanks for reporting the bug. Could you please try "make -k check"
to see if there are any more like it?
However, I don't think this bug indicates a serious problem with
coreutils; it's just the usual MacOS glitch with file name case
sensitivity.