>>>> "Jarl" == Jarl Friis <jarl(a)diku.dk>
Jarl> On 17 Dec 2001, Adrian Aichner wrote:
> >>>>> "SY" == Steve Youngs
SY> |--==> "n" == nbecker <nbecker(a)fred.net>
> >>>>>>> "SY" == Steve Youngs
n> I marked a bunch of files in pcl-cvs
mode. Then I chose "undo
n> changes" from the menu. It prompts "Delete 8 files?". Rather
SY> Actually, this is correct behaviour. The menu
item "undo changes"
SY> runs 'cvs-mode-undo', that function undoes all the local changes to
SY> the marked files in the PCL-CVS buffer. It does it by deleting the
SY> local file and then runs 'cvs update FILE'.
n> It may be literally correct, but I was surprised
and confused by this
n> question. I wasn't quite sure whether it was doing just what you
n> describe, or whether it had gone insane and was offering to delete 8
n> files at random. IIRC, it didn't tell me which 8 files it wanted to
n> delete, which didn't help either.
SY> Do you think it would be better and less
confusing if you were
SY> prompted with something like "Delete local copy of marked
SY> files and re check them out of CVS?"? Or words to that
>> Hi Steve,
>> I hate the current behavior for the reasons
>> It's confusing.
>> It does not provide any information about the files affected.
Jarl> What about make it look similar to when committing, except
Jarl> the message entering part. After entering a message and hit
Jarl> C-c C-c a small buffer called *cvs-edit-files* comes up with
Jarl> a list of the files affected and you are prompted "Really
Jarl> commit? (y or n)", similarly cvs-mode-undo could come up
Jarl> with a *cvs-edit-files* buffer listing the files affected,
Jarl> and ask something like "Really undo ? (y or n)" or "Really
Jarl> revert from CVS? (y or n)" I suggest *not* to use "Really
Jarl> delete ? (y or n)" for reasons given below.
This is basically what I called "completion buffer feedback".
Jarl> In both cases the prompt could be improoved to contain the number of
Jarl> files, such as "Really commit 8 files (y or n)?".
>> I think completion buffer feedback, along the
lines of what
>> d runs the command dired-flag-file-deletion
>> x runs the command dired-expunge-deletions
>> do, would be much better.
Jarl> I am not sure I get you right, I guess you want behaviour
Jarl> similar to deletion of files (in `dired') and buffers (in
I pointed out an example of the UI I have in mind.
I don't want to know whether 894 or 895 files are going to be
reverted, I wnat to see what their names are.
Jarl> I don't think it's a good idea to think of the cvs-mode-undo
Jarl> as a *deletion*, because the files are still there after a
I don't think that either, that's why I suggested:
(defun-cvs-mode (cvs-mode-undo . SIMPLE) (flags)
in pcl-cvs.el I would reword the message to:
"Undo local changes?"
If that message is no good, then the docstring of cvs-mode-undo should
be changed also, that's my point.
Jarl> cvs-mode-undo the content has just changed. I know that the
Jarl> effect is delete-then-update, but from a users point of view
Jarl> I think it more looks like an undo or a revert.
Jarl> and "yes" PCL-CVS is great!