This is the mail archive of the
mailing list for the Cygwin project.
Re: [PATCH]: Still stripping
On Thu, Dec 23, 2004 at 02:51:34PM -0500, Pierre A. Humblet wrote:
>Christopher Faylor wrote:
>> On Thu, Dec 23, 2004 at 11:42:15AM -0500, Pierre A. Humblet wrote:
>> >In a case such as "abc..exe", the posix_path "abc." should not be
>> >stripped. The patch below only strips the posix path if the win32
>> >path was stripped. I don't think that the posix path can be empty
>> >in that case.
>> >2004-12-23 Pierre Humblet <email@example.com>
>> > * path.h (path_conv::set_normalized_path): Add second argument.
>> > * path.cc (path_conv::check): Declare, set and use "strip_tail".
>> > (path_conv::set_normalized_path): Add and use second argument,
>> > replacing all tail stripping tests.
>> I'm not sure that your assumption of dot stripping is true in the first
>> case of set_normalized_path in build_fh_dev in dtable.cc.
>Not sure I understand what you mean. At any rate there are two cases
>where build_fh_dev is called with a non-empty second argument:
>handler_tty.cc: console = (fhandler_console *) build_fh_dev
>path.cc: fhandler_virtual *fh = (fhandler_virtual *) build_fh_dev (dev, path_copy);
>Neither is about a disk path.
True, but there was at least a brief time when "ls /proc/cpuinfo."
worked. So this would be a reversion from the past two releases of
cygwin. Maybe it doesn't matter and, it seems like generically
stripping paths in build_fh_dev will not work anyway. So, I guess we'll
just have to make sure that people are aware of this reversion.