Re: Bug in canonicalize_path()

I have test the following on Windows and Linux:

Windows:
C:\>cd Winnt

C:\WINNT>cd C:\Temp\..\..\

C:\>

Linux:
$ cd /usr/../../
$ pwd
/

We should handle this correctly.

1 Single dot in the path can be removed safety. (except the first one. e.g. ./usr/local)
2 Every double dot may need a removal of the last part of the path. (except the first one. e.g. ../local)
And if there are not enough part left, keep the last part as it is.

We can even make it easier by adding step 0: make sure path is an absolute path.

----- Original Message -----
From: "Bruce Momjian" <pgman(at)candle(dot)pha(dot)pa(dot)us>
To: "Tom Lane" <tgl(at)sss(dot)pgh(dot)pa(dot)us>
Cc: "William ZHANG" <uniware(at)zedware(dot)org>; <pgsql-patches(at)postgresql(dot)org>
Sent: Friday, August 12, 2005 11:15 AM
Subject: Re: [PATCHES] Bug in canonicalize_path()

> Tom Lane wrote:
> > Bruce Momjian <pgman(at)candle(dot)pha(dot)pa(dot)us> writes:
> > > But what about "usr/local/../../.."?
> >
> > What about it? The case of /usr/local/../../.. is handled correctly,
> > and the case where it's an underspecified relative path doesn't seem
> > that interesting to me --- certainly that is not so important that we
> > should get the wrong answer on cases that *are* plausible.
> >
> > Most of the uses of canonicalize_path are on paths that are required to
> > be absolute, anyway.
> >
> > It wouldn't be too implausible to error out if pending_strips>0 after
> > exiting the loop.
>
> I figured it would be best to leave it alone if we can't process it, but
> if you think it is more imporant to trim in cases like ../.., go ahead.
>
> --
> Bruce Momjian | http://candle.pha.pa.us
> pgman(at)candle(dot)pha(dot)pa(dot)us | (610) 359-1001
> + If your life is a hard drive, | 13 Roberts Road
> + Christ can be your backup. | Newtown Square, Pennsylvania 19073