On Wednesday 07 December 2005 14:37, Gale, David wrote:
> Erik Huelsmann wrote:
[snip]
> > THEN FIX IT!
>
> Erik, I must say that this isn't a good response. Gary has posted twice
> in this thread, both times complaining that the attitude of "this is the
> way things work right now, so it's not a bug" is annoying &
> unbeneficial--a sentiment which I happen to agree with. Neither of
> Gary's posts have indicated that he is a programmer, or that he can
> spend several hours a night (thank you for your willingness to do that,
> by the way) working on fixing this behavior. Sniping at him like this
> is extremely unprofessional.
Look, nobody is going to disagree about the desired end goal of a rename and
edit operation, when the original file has been modified in the repository.
The Subversion team wants to fix that issue. However, it is wrought with
complexities, and edge cases. Which is why a minor release is geared towards
solving the single biggest problem of describing a rename operation in the
repository.
On Erik's behalf, he's put quite a few hours into solving some very real
problems in Subversion (I honestly don't know where he finds the time... he's
basically working a second job and not being paid for it). At times, the
general tone of this list changes to one where most posters forget that
people volunteer their time and try to dictate how we should spend our time
solving their particular issue as though it's the only thing we have going on
in our lives. It's annoying and stressful, and after some quantity of time,
results in lashing out at someone who has good intentions. Erik is a really
nice guy, please don't hold this one experience against him.
I also think there was some miscommunication about the difference between a
bug and what the expected behavior was from a user's perspective, and whether
or not this particular situation was the former. Greg, I appreciate the fact
that you pointed this out. Yes, there exists more than one type of "bug",
and this definitely falls into the requirements category. I believe we're
over that hump, so let's move on.
> Better would be to be to point out Issue 898, "implement true renames",
> which (if fixed) would actually change the current behavior to match
> what has been asked for in this thread. Since Issue 898 is classified
> as a "DEFECT" rather than an "ENHANCEMENT" or "FEATURE", it would appear
> that the dev. community agrees that this is, in fact, a bug. Also,
> there are (currently) 50 votes for this issue--the second highest issue
> by votes in the Issue Tracker, making it fairly clear that the user
> community also feels fairly strongly that this is a major issue. Of
> course, reading through the comments for 898 make it clear that fixing
> this will be a fairly involved process.
I wouldn't go as far as saying that. Issue 898 is talking about removing the
first part of the problem: how to describe the rename properly in the
repository. Half the people that voted for that issue probably don't
understand that it has nothing to do with what would happen in someone's
working copy. Issue 2282 is the one talking about how to handle tree edits
better (which is marked as an enhancement), and their impact on the working
copy.
BTW, we don't use the votes in the Issue Tracker for anything. We count on
user feedback to help determine the things we need to fix. Given this
thread, I'd say that this is yet another group of users who believe true
renames and better tree edit handling are very important to them.
If it helps at all, I certainly feel like these have been a long-standing
issues that needs to be addressed. I think there has been a lot of
resistance just because of the underlying complexity of solving it. It's
going to require modifications at nearly every layer of Subversion to solve
(fs, repository, wc, and ra), and therefore requires a lot of knowledge and
coordination with other developers in order to generate a complete and
comprehensive solution. In general, the community is very good about digging
deep and attacking the underlying issues of the problem. I think the current
move semantics is probably one spot where we compromised functionality in
order to move forward with other features. Hindsight is 20/20. :-)
Let's work on getting 1.3.0 wrapped up, so that some new things can happen on
trunk/ for 1.4. Hopefully, both of these issues (representing renames in the
repo, and the corresponding actions in the WC) will get solved then.
-John
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@subversion.tigris.org
For additional commands, e-mail: users-help@subversion.tigris.org
Received on Thu Dec 8 11:55:28 2005