Re: [Monotone-devel] xmtn: Emacs integration via DVC

[Thomas Moschny]

Christian Ohler wrote:
> Thomas Moschny, 2007-04-12:
> > On Tuesday 03 April 2007, Christian Ohler wrote:
> >> Pressing C-x V = will bring up the tree diff buffer.  (What monotone
> >> calls a "workspace" is called a "tree" in DVC.)  This buffer shows the
> >> list of all modified files in the tree as well as the diffs for those
> >> files.  Use j to jump back and forth between the name of a file in the
> >> list and the diffs for that file.  Use RET with point inside a diff
> >> hunk to go to the corresponding file at the corresponding position.
> >
> > Using the j key works fine. However, the highlighting suggests that
> > jumping to the hunk currently under the mouse would be possible using
> > mouse-2, but pressing the middle button over the (highlighted) file name
> > gives me a "diff-beginning-of-hunk: Can't find the beginning of the hunk"
> > message.
>
> I'm afraid I can't reproduce this.  My mouse only has buttons 1 and 3.
>
> :)
>
> More seriously, you're clicking on one of the highlighted (blue for me)
> filenames near the top of the buffer, and Emacs gives you that error?

In the "*xmtn-diff*" buffer, when I move the mouse over the list of changed 
files at the top (after the revision sniplet in basic_io format, and before 
the diffs), each file gets highlighted, and a yellow bubble pops up. It is 
labeled "File Changes" and gives a list of keys and bindings. It says that 
<mouse-2> is bound to dvc-diff-jump-to-change-by-mouse. However, C-h k tells 
me that <mouse-2> is bound to duff-mouse-goto-source at that spot.

Now, when I press the middle mouse button while being over one of the files, I 
get "diff-beginning-of-hunk: Can't find the beginning of the hunk".

> Does the same thing happen when you move point onto that filename and hit
> RET?

No, this correctly jumps into the buffer for that file.

> Or are you clicking on a filename in a diff header?

No, but this yields the same error, at least for the first hunk. However, and 
that may give you a hint on where the error lays: When I middle-click on the 
filename in the header of the _second_ hunk, it jumps to the buffer for the 
filename of the _first_ hunk. Same for RET while point is in the diff header.

So, there must be something wrong with the formatting and/or parsing of the 
*xmtn-diff* buffer.

> (Are they getting highlighted on your system?  Not on mine...)  

They are not highlighted (besides they are constantly having a different face 
than the rest of the diff).

> Could you post a backtrace of that error, please?  (M-x
> toggle-debug-on-error) 

Here's the backtrace for a middle click on one of the filenames in the list:

Debugger entered--Lisp error: (error "Can't find the beginning of the hunk")
  signal(error ("Can't find the beginning of the hunk"))
  error("Can't find the beginning of the hunk")
  diff-beginning-of-hunk()
  diff-find-source-location(nil t)
  diff-goto-source()
  diff-mouse-goto-source((mouse-2 (#<window 7 on *xmtn-diff*> 314 (45 . 163) 
25166195)))
* call-interactively(diff-mouse-goto-source)


Regards,
Thomas

pgpqibOoPrQrx.pgp
Description: PGP signature