Re: Weird scoping issue in options file

On Thu, Aug 20, 2009 at 6:52 PM, Michael Haggerty<[email protected]> wrote:
> I think that your fix has nothing to do with where ctx and run_options
> are stored.  Rather it has to do with the fact that you pass only one
> dictionary to execfile(), which means that a single dictionary is used
> for both globals and locals.  Therefore the file-level imports end up
> (also) in the globals dictionary, and are visible from within nested scopes.

Hmmm, OK, I'll buy that.

> I have committed your suggested patch (as trunk r4879), since it seems
> like the right thing to do (even aside from the fact that it actually
> works).

Cool.  Is there any way to write a test for that?  It would be nice to
catch if someone in 3 years breaks the execfile() call.  The fact that
you couldn't repro my problem with a cvs2svn options file is
troubling.

Why don't I just explain exactly what I did.  First, I turned
cvs2svn-example.options into cvs2svn-test.options with this patch:

"""
--- cvs2svn-example.options	2009-08-18 15:53:37.000000000 -0400
+++ cvs2svn-test.options	2009-08-21 09:32:34.000000000 -0400
@@ -130,6 +130,12 @@
 # cvs2svn conversion.  Please choose one of the following and adjust
 # the parameters as necessary:

+class MyOutputOption(NewRepositoryOutputOption):
+    def __init__(self, *args, **kwargs):
+        #from cvs2svn_lib.svn_output_option import NewRepositoryOutputOption
+        NewRepositoryOutputOption.__init__(self, *args, **kwargs)
+
+
 # Use this output option if you would like cvs2svn to create a new SVN
 # repository and store the converted repository there.  The first
 # argument is the path to which the repository should be written (this
@@ -141,7 +147,7 @@
 # options to be passed to "svnadmin create".  The (optional)
 # author_transforms argument allows CVS author names to be transformed
 # arbitrarily into SVN author names (as described above):
-ctx.output_option = NewRepositoryOutputOption(
+ctx.output_option = MyOutputOption(
     r'/path/to/svnrepo',
     #fs_type='fsfs',
     #bdb_txn_nosync=False,
"""

(Note the commented-out import at method scope, necessary for this to
work pre-r4879.)

Then I ran

  cvs2svn --options cvs2svn-test.options

And it crashed pretty darn soon.

Is that enough info to create a proper test case?

Greg

------------------------------------------------------
http://cvs2svn.tigris.org/ds/viewMessage.do?dsForumId=1667&dsMessageId=2386000

To unsubscribe from this discussion, e-mail: [[email protected]].