should `G r' rename score files?

should `G r' rename score files?

[Dan Christensen]

When I rename an nnfolder group using `G r' in the *Group* buffer,
this doesn't rename the .SCORE and .ADAPT files.  Maybe it should?

Dan

Re: should `G r' rename score files?

[Kevin Greiner]

I considered implementing this several months ago.  I ended up
abandoning the effort because of the way that hierarchical score files
function.  I'll use an example to describe the problem.

Let's say that you have the groups:
a.b.c
a.b.d

and the score files:
a.b.c.SCORE
a.b.d.SCORE
a.b.SCORE
a.SCORE

Now, if I rename a.b.c to x.y.z.  What do I do about the score files?

Renaming a.b.c.SCORE to x.y.z.SCORE is obvious.  The problem is what
to do with a.b.SCORE and a.SCORE.

The options are:
* If I do nothing, then the score rules for x.y.z will not match the
  score rules used when the group was named a.b.c.
* If I rename a.b.SCORE and a.SCORE, then the score rules for a.b.d
  are changed.
* If I copy a.b.SCORE to x.y.SCORE and a.SCORE to x.SCORE, then these
  groups are OK but what if x.SCORE or x.y.SCORE already existed?

To make this work, I'd have to merge the a.b.SCORE contents with
x.y.SCORE in such a way that I restored the proper data structure,
removing all duplicate entries, and (if possible) preserving any
comments.  It's possible but a lot of work to get right.

Kevin

Dan Christensen <jdc@uwo.ca> writes:

> When I rename an nnfolder group using `G r' in the *Group* buffer,
> this doesn't rename the .SCORE and .ADAPT files.  Maybe it should?

Re: should `G r' rename score files?

[Ted Zlatanov]

On Wed, 15 Sep 2004, kevin.greiner@vignette.com wrote:

> I considered implementing this several months ago.  I ended up
> abandoning the effort because of the way that hierarchical score files
> function.  I'll use an example to describe the problem.
> 
> Let's say that you have the groups:
> a.b.c
> a.b.d
> 
> and the score files:
> a.b.c.SCORE
> a.b.d.SCORE
> a.b.SCORE
> a.SCORE
> 
> Now, if I rename a.b.c to x.y.z.  What do I do about the score files?
> 
> Renaming a.b.c.SCORE to x.y.z.SCORE is obvious.  The problem is what
> to do with a.b.SCORE and a.SCORE.
> 
> The options are:
> * If I do nothing, then the score rules for x.y.z will not match the
>   score rules used when the group was named a.b.c.
> * If I rename a.b.SCORE and a.SCORE, then the score rules for a.b.d
>   are changed.
> * If I copy a.b.SCORE to x.y.SCORE and a.SCORE to x.SCORE, then these groups
> 	are OK but what if x.SCORE or x.y.SCORE already existed?

I think the user should be alerted at the time of renaming:

"The score files a.b and a will no longer apply to group x.y.z
(formerly a.b.c).  Proceed with group rename? (y/n)"

(abort the rename if necessary) and then

"The score file a.b.c should be renamed to x.y.z.  Proceed with score
file rename? (y/n)"

I think that gives the user a chance to do the right thing or at least
know what will be affected.

Ted

Re: should `G r' rename score files?

[Kai Grossjohann]

Kevin Greiner <kevin.greiner@vignette.com> writes:

> Renaming a.b.c.SCORE to x.y.z.SCORE is obvious.  The problem is what
> to do with a.b.SCORE and a.SCORE.
>
> The options are:
> * If I do nothing, then the score rules for x.y.z will not match the
>   score rules used when the group was named a.b.c.

I think this is the best solution.  It is obvious what it does.  If
you tried to take a.b.SCORE and a.SCORE into account, then the next
step would be to also take gnus-parameters into account, and so
forth.  Down that path lies madness.

After all, the hierarchical score file stuff is defined to work on the
group name, so it is not surprising that a different group name leads
to different score rules.

However, what you should do when renaming the gruop a.b.c to a.b, I
don't know.  Perhaps just signal an error explaining the situation.

Kai