[svn.haxx.se] · SVN Dev · SVN Users · SVN Org · TSVN Dev · TSVN Users · Subclipse Dev · Subclipse Users · this month's index

Re: svn commit: r10133 - in trunk/subversion: include libsvn_subr tests/clients/cmdline tests/libsvn_subr

From: Greg Hudson <ghudson_at_MIT.EDU>
Date: 2004-07-04 17:13:01 CEST

On Sun, 2004-07-04 at 09:01, jpieper@tigris.org wrote:
> Issue #1779: Improve svn_path_canonicalize to make more
> transformations aimed at removing path specification redundancies.

I am concerned about the case where one runs something like "svn st
foo/../bar", where foo is a symlink with multiple path elements.
(Symlinks within working dirs around allowed, but I see no reason why
symlinks should be disallowed in the portion of a path leading up to the
working copy.)

You did not break this case; it was already broken due to the call to
apr_filepath_merge() in libsvn_subr/opt.c:svn_opt_args_to_target_array.
But it seems like the logic of both apr_filepath_merge() and the new
logic of svn_path_canonicalize() are fundamentally flawed, unless they
are only applied within a playground where directory symlinks are
disallowed.

> (svn_path_canonicalize): Replaced old implementation with new that
> uses nearly identical logic to large portions of
> apr_filepath_merge. It condenses redundant "/./", "/../", and
> "//" portions of the path, as well as removing any trailing
> separator characters.

When you say in a log message that a function is "nearly identical" to a
preexisting function, you might say how it's different. As it stands,
I'm not sure why you didn't use apr_filepath_merge to implement
svn_path_canonicalize.

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Sun Jul 4 17:13:40 2004

This is an archived mail posted to the Subversion Dev mailing list.