Attachment fail on Win32 without normalization of env.path

[Shun-ichi Goto <gotoh@…>]

I think that the value of Environment.path should be normalized nortation like 'c:\trac\project' not 'c:/trac/project' on windows environment. This style like 'c:/dir/dir' is usual in httpd.conf but it would cuase unexpected result of os.path.commonprefix() used in attachment.py. Thus adding attachement would fail.

Index: env.py
===================================================================
--- env.py      (revision 1846)
+++ env.py      (working copy)
@@ -78,7 +78,7 @@
         except ImportError:
             pass

-        self.path = path
+        self.path = os.path.normpath(path)
         self.__cnx_pool = None
         if create:
             self.create(db_str)

[Matthew Good]

#1719 has been marked as a duplicate of this ticket.

[PBruin]

Maybe a TracError would be usefull. I think many people get shock if their configuration that was working fine for month stops working after doing an update.

Index: attachment.py
===================================================================
--- attachment.py	(revision 1862)
+++ attachment.py	(working copy)
@@ -122,7 +122,10 @@
         # attachments directory
         attachments_dir = os.path.join(self.env.path, 'attachments')
         commonprefix = os.path.commonprefix([attachments_dir, self.path])
-        assert commonprefix == attachments_dir
+        if not commonprefix == attachments_dir:
+           raise TracError('Common prefix (%s) of attachments_dir (%s) and self.path (%s) not ok' %
+                           (commonprefix,attachments_dir,self.path),
+                           'Attachment failed. Check TracEnv!' );
 
         if not os.access(self.path, os.F_OK):
             os.makedirs(self.path)

[Christian Boos]

I had to make a quick fix for this, in preparation for other changes in attachment.py. I didn't modify env.py as suggested by the reporter, but instead fixed that particular failure in attachment.py: That way, it's less likely to introduce unwanted side-effects…

(see [1865])