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

Re: Current Testing Situation

From: Ben Collins-Sussman <sussman_at_collab.net>
Date: 2001-04-26 15:59:58 CEST

Sam.... you need to relax. Remember my commit said "this isn't done
yet. Please don't try to use it."

All of these problems come from the fact that this is a work in
progress. I did the commit for safety's sake and so we could start a
discussion. There's nothing broken because there's nothing to break
-- we have no python testsuite (yet). :)

Please don't try to patch something that's a moving target. Instead,
let's talk about the design.

Sam TH <sam@uchicago.edu> writes:
 
> [This is mostly for Ben, but the rest of you might care as well.]
>
> There are a number of problems that are currently causing the entire
> python testsuite to fail.
>
> 1. xml_tests has not been updated to reflect either the changed
> locations of some functions (like svn_wc vs svn_tree), the new names
> (wc_to_tree vs build_tree_from_wc), or the new return values. These are
> just mechanical changes.
>
> 2. Either you neglected to commit, or just didn't like, the code to
> convert entries files to a tree. This naturally makes using such code
> hard.
>
> 3. The expected output for test 2 is a bunch of regular expressions,
> which do not translate nicely to a tree.
>
> 4. The comparison function is not smart enough to understand that some
> nodes lack of content is not a statement about the content, but a
> statement about where the node came from. Therefore, it doesn't like
> comparing a node from output paths to a node from a real file. It's
> possible to hack around this by extending compare_trees so that it can
> be told not to look at contents or props. But this should be done
> automatically.
>
> 5. A number of the tree creation functions that you wrote
> (tree_from_checkout, wc_to_tree) include the working copy directory,
> called something like wc-t1, in the tree structure. This is a
> mistake, since the name of that directory is arbitrary and not
> controlled by subversion. Also, this makes it hard to compare trees
> from two different working copies. I managed to kludge up your
> wc_to_tree function to skip this dir, but didn't know how to for
> tree_from_checkout.
>
> Working around problems 1 and 2 was easy. Problem 3 I skipped, since
> it relates to a more difficult issue I brought up in my last mail. I
> kludged solutions to parts of 4 and 5. This got me to the point where
> 3 of the 4 tests pass. However, better solutions need to be found for
> 3, 4 and 5.
>
> Ideas?
>
> sam th --- sam_at_uchicago.edu --- http://www.abisource.com/~sam/
> OpenPGP Key: CABD33FC --- http://samth.dyndns.org/key
> DeCSS: http://samth.dynds.org/decss
>
> Content-Type: application/pgp-signature
> Content-Disposition: inline
>
>
> This message contains data in an unrecognized format, application/pgp-signature,
> which is being decoded and written to the file named "/tmp/mm.q40629".
> If you do not want this data, you probably should delete that file.
> Wrote file /tmp/mm.q40629
Received on Sat Oct 21 14:36:29 2006

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