Bug: Regressions from 8.2.10 (8.2.10-35-g19a7d6-elpaplus) to master/maint

Bug: Regressions from 8.2.10 (8.2.10-35-g19a7d6-elpaplus) to master/maint

[Jan Malakhovski]

[-- Attachment #1: Type: text/plain, Size: 1029 bytes --]

Hi.

Emacs           : GNU Emacs 24.5.2 (x86_64-unknown-linux-gnu, GTK+ Version 2.24.28) of 2015-10-05 on localhost
Current  package: Org-mode version 8.3.2 (release_8.3.2-209-gba4d33 @ /home/oxij/repo/org-mode/lisp/) (actually, both master and maint)
Previous package: Org-mode version 8.2.10 (8.2.10-35-g19a7d6-elpaplus @ /run/current-system/sw/share/emacs/site-lisp/elpa/org-plus-contrib-20150406/)

Some regressions when moving from previous to current package:

* C-c C-x C-i clocks in into a wrong node.
  Instead of clocking-in into the node under cursor it clock-ins into the
  next one.
  Can be replicated with the following init.el:

#+BEGIN_SRC emacs-lisp
(add-to-list 'load-path "~/repo/org-mode/lisp")
(add-to-list 'load-path "~/repo/org-mode/contrib/lisp")
(require 'org)
#+END_SRC

  and attached minimal.org.

* Pressing <Tab> on an entry in org-agenda-list moves the cursor to the
  corresponding node.

  Previously it moved the cursor to the corresponding CLOCK: line and I
  used that a lot.

Cheers,
  Jan


[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #2: minimal.org --]
[-- Type: text/x-org, Size: 47 bytes --]

* Test
** Test test
** Test test
* Test
* Test

Re: Bug: Regressions from 8.2.10 (8.2.10-35-g19a7d6-elpaplus) to master/maint

[Nicolas Goaziou]

Hello,

Jan Malakhovski <oxij@oxij.org> writes:

> * C-c C-x C-i clocks in into a wrong node.
>   Instead of clocking-in into the node under cursor it clock-ins into the
>   next one.

Fixed. Thank you.

> * Pressing <Tab> on an entry in org-agenda-list moves the cursor to the
>   corresponding node.
>
>   Previously it moved the cursor to the corresponding CLOCK: line and I
>   used that a lot.

According to `org-agenda-goto''s docstring, this doesn't look like
a feature. Besides, there is a function to jump to the running clock
already: `org-clock-goto'.


Regards,

-- 
Nicolas Goaziou

Re: Bug: Regressions from 8.2.10 (8.2.10-35-g19a7d6-elpaplus) to master/maint

[Jan Malakhovski]

Nicolas Goaziou <mail@nicolasgoaziou.fr> writes:

>> * Pressing <Tab> on an entry in org-agenda-list moves the cursor to the
>>   corresponding node.
>>
>>   Previously it moved the cursor to the corresponding CLOCK: line and I
>>   used that a lot.
>
> According to `org-agenda-goto''s docstring, this doesn't look like
> a feature. Besides, there is a function to jump to the running clock
> already: `org-clock-goto'.

I mean this sequence:

* Press `M-x org-agenda-list`
* Press `v l`
  (or `v c`)
* Select an entry (interval)
* Press `<Tab>`

Previously it opened a window with the cursor on the CLOCK: line of the
corresponding entry (interval), now it jumps to the header of the task
this interval is an element of.

I'm sure this is a misfeature because now it's pretty much impossible to
quickly find CLOCK: entries that correspond to current org-agenda-list
items. For me this is critical, because I clock everything, have
megabytes of CLOCK: entries and edit them a lot.

Cheers,
  Jan

Re: Bug: Regressions from 8.2.10 (8.2.10-35-g19a7d6-elpaplus) to master/maint

[Nicolas Goaziou]

Jan Malakhovski <oxij@oxij.org> writes:

> Nicolas Goaziou <mail@nicolasgoaziou.fr> writes:
>
>>> * Pressing <Tab> on an entry in org-agenda-list moves the cursor to the
>>>   corresponding node.
>>>
>>>   Previously it moved the cursor to the corresponding CLOCK: line and I
>>>   used that a lot.
>>
>> According to `org-agenda-goto''s docstring, this doesn't look like
>> a feature. Besides, there is a function to jump to the running clock
>> already: `org-clock-goto'.
>
> I mean this sequence:
>
> * Press `M-x org-agenda-list`
> * Press `v l`
>   (or `v c`)
> * Select an entry (interval)
> * Press `<Tab>`
>
> Previously it opened a window with the cursor on the CLOCK: line of the
> corresponding entry (interval), now it jumps to the header of the task
> this interval is an element of.

Could you provide an ECM with appropriate CLOCK lines?

Thank you.


Regards,

Re: Bug: Regressions from 8.2.10 (8.2.10-35-g19a7d6-elpaplus) to master/maint

[Jan Malakhovski]

[-- Attachment #1: Type: text/plain, Size: 676 bytes --]

Nicolas Goaziou <mail@nicolasgoaziou.fr> writes:

> Could you provide an ECM with appropriate CLOCK lines?

Attached.

* Other files:
** init.el:

#+BEGIN_SRC emacs-lisp
(add-to-list 'load-path "~/repo/org-mode/lisp")
(add-to-list 'load-path "~/repo/org-mode/contrib/lisp")
(require 'org)

(setq org-directory "~/org"
      org-agenda-files "~/agenda-files")
#+END_SRC

** ~/agenda-files:

#+BEGIN_SRC emacs-lisp
~/org/minimal.org
#+END_SRC

* Process:
** Press `M-x org-agenda-list`
** Press `v l` (or `v c`)
** Select any of the two entries (intervals)
** Press `<Tab>`
* Expected: cursor jump to the corresponding CLOCK:.
* Got: cursor jump to "Test" line.

Cheers,
  Jan


[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #2: minimal.org --]
[-- Type: text/x-org, Size: 149 bytes --]

* Test
:LOGBOOK:
CLOCK: [2015-10-25 Sun 17:00]--[2015-10-25 Sun 18:00] =>  1:00
CLOCK: [2015-10-25 Sun 16:00]--[2015-10-25 Sun 17:00] =>  1:00
:END:

Re: Bug: Regressions from 8.2.10 (8.2.10-35-g19a7d6-elpaplus) to master/maint

[Jan Malakhovski]

Jan Malakhovski <oxij@oxij.org> writes:

> Nicolas Goaziou <mail@nicolasgoaziou.fr> writes:
>
>> Could you provide an ECM with appropriate CLOCK lines?
>
> Attached.

Okay, so I bisected and found that the offender is
9f5e698679aecbed872a2030e4157e5e2b1d87e0. Since then these lines have
changed, so the following

diff --git a/lisp/org-agenda.el b/lisp/org-agenda.el
index bdb69c5..38baac2 100644
--- a/lisp/org-agenda.el
+++ b/lisp/org-agenda.el
@@ -8411,10 +8411,10 @@ When called with a prefix argument, include all archive files as well."
 	     (org-flag-heading nil)))	; show the next heading
       (when (outline-invisible-p)
 	(outline-show-entry))			; display invisible text
-      (recenter (/ (window-height) 2))
-      (org-back-to-heading t)
-      (if (re-search-forward org-complex-heading-regexp nil t)
-	  (goto-char (match-beginning 4))))
+      (recenter (/ (window-height) 2)))
+;      (org-back-to-heading t)
+;      (if (re-search-forward org-complex-heading-regexp nil t)
+;	  (goto-char (match-beginning 4))))
     (run-hooks 'org-agenda-after-show-hook)
     (and highlight (org-highlight (point-at-bol) (point-at-eol)))))

makes all of my problems this go away.

If you read 9f5e698679aecbed872a2030e4157e5e2b1d87e0 you'll notice that
it changed both the behavior and the doc-string (that is to say that you
can't refer to doc-string as an argument against the previous behavior).

I'm pretty sure that there are other org-mode users that are weeping
silently since that change. Suppose you have more than a screen of clock
lines in LOGBOOK under a heading (I have headings that have thousands).
(org-agenda-goto) before 9f5e698679aecbed872a2030e4157e5e2b1d87e0 is
very helpful when moving CLOCK lines between headings (think "refile
CLOCK line") or editing overlaps or gaps: you <Tab> or click on the
heading and here you are at the CLOCK, but after
9f5e698679aecbed872a2030e4157e5e2b1d87e0 you have to *retype* the date
and time into search, which is annoying to say the least.

I'm willing to listen, but I seriously doubt there's such a compelling
argument defending the change. Even when out of `org-agenda-list` in
plain `org-agenda`, jumping to SCHEDULED or DEADLINE lines is nicer than
to the heading itself.

Cheers,
  Jan

Re: Bug: Regressions from 8.2.10 (8.2.10-35-g19a7d6-elpaplus) to master/maint

[Nicolas Goaziou]

Jan Malakhovski <oxij@oxij.org> writes:

> Okay, so I bisected and found that the offender is
> 9f5e698679aecbed872a2030e4157e5e2b1d87e0. Since then these lines have
> changed, so the following

[...]

> I'm willing to listen, but I seriously doubt there's such a compelling
> argument defending the change. Even when out of `org-agenda-list` in
> plain `org-agenda`, jumping to SCHEDULED or DEADLINE lines is nicer than
> to the heading itself.

I don't know either why that was changed, but there probably was
a reason for it. Bastien may remember it.

Regards,

Re: Bug: Regressions from 8.2.10 (8.2.10-35-g19a7d6-elpaplus) to master/maint

[Matt Lundin]

Jan Malakhovski <oxij@oxij.org> writes:

> I'm pretty sure that there are other org-mode users that are weeping
> silently since that change. Suppose you have more than a screen of clock
> lines in LOGBOOK under a heading (I have headings that have thousands).
> (org-agenda-goto) before 9f5e698679aecbed872a2030e4157e5e2b1d87e0 is
> very helpful when moving CLOCK lines between headings (think "refile
> CLOCK line") or editing overlaps or gaps: you <Tab> or click on the
> heading and here you are at the CLOCK, but after
> 9f5e698679aecbed872a2030e4157e5e2b1d87e0 you have to *retype* the date
> and time into search, which is annoying to say the least.

Try typing RET in the agenda. That calls org-agenda-switch-to, which
brings you to the relevant clock information in the agenda. (However, if
the information is tucked away in a LOGBOOK, it remains somewhat
inconveniently hidden from view).

> I'm willing to listen, but I seriously doubt there's such a compelling
> argument defending the change. Even when out of `org-agenda-list` in
> plain `org-agenda`, jumping to SCHEDULED or DEADLINE lines is nicer than
> to the heading itself.

I disagree that jumping to the SCHEDULED or DEADLINE lines from a normal
agenda is always nicer than jumping to the headline. I can easily change
planning info from within the agenda. But I cannot change the text of
the headline itself -- that is usually why I jump from the agenda. If
nothing else, we should have both options, which we seem to currently
have with the difference between TAB (org-agenda-goto) and RET
(org-agenda-switch-to). However, I am not sure how intentional this is,
as I can find nothing highlighting this difference in the documentation.

Best,
Matt

Re: Bug: Regressions from 8.2.10 (8.2.10-35-g19a7d6-elpaplus) to master/maint

[Jan Malakhovski]

Matt Lundin <mdl@imapmail.org> writes:

> Jan Malakhovski <oxij@oxij.org> writes:
>
>> I'm pretty sure that there are other org-mode users that are weeping
>> silently since that change. Suppose you have more than a screen of clock
>> lines in LOGBOOK under a heading (I have headings that have thousands).
>> (org-agenda-goto) before 9f5e698679aecbed872a2030e4157e5e2b1d87e0 is
>> very helpful when moving CLOCK lines between headings (think "refile
>> CLOCK line") or editing overlaps or gaps: you <Tab> or click on the
>> heading and here you are at the CLOCK, but after
>> 9f5e698679aecbed872a2030e4157e5e2b1d87e0 you have to *retype* the date
>> and time into search, which is annoying to say the least.
>
> Try typing RET in the agenda. That calls org-agenda-switch-to, which
> brings you to the relevant clock information in the agenda.(However, if
> the information is tucked away in a LOGBOOK, it remains somewhat
> inconveniently hidden from view).

Judging by the names and default key bindings alone I'd prefer
org-agenda-switch-to to do the new org-agenda-goto thing, and
org-agenda-goto have the previous behavior. This kinda makes more sense
to me, see below.

In any case, current org-agenda-switch-to behavior is no replacement for
old org-agenda-goto:

1) I'm not aware of a key sequence to do org-reveal for drawers (in
org-agenda-goto it's a nontrivial piece of code).
2) Even if there is such a sequence, the drawer should be open by default.

>> I'm willing to listen, but I seriously doubt there's such a compelling
>> argument defending the change. Even when out of `org-agenda-list` in
>> plain `org-agenda`, jumping to SCHEDULED or DEADLINE lines is nicer than
>> to the heading itself.
>
> I disagree that jumping to the SCHEDULED or DEADLINE lines from a normal
> agenda is always nicer than jumping to the headline. I can easily change
> planning info from within the agenda. But I cannot change the text of
> the headline itself -- that is usually why I jump from the agenda. If
> nothing else, we should have both options, which we seem to currently
> have with the difference between TAB (org-agenda-goto) and RET
> (org-agenda-switch-to). However, I am not sure how intentional this is,
> as I can find nothing highlighting this difference in the documentation.

Okay, that is a pretty valid use case, but even then, (below starts
here) org file syntax requires SCHEDULED and DEADLINE to be just under
the heading, which means that you can navigate to the heading pretty
easily, where as with CLOCK lines you can not. Which is why, I think,
both org-agenda-goto and org-agenda-goto-mouse should have the old
behavior, but, in principle, I can imagine <Tab> and <RET> doing
different things in plain org-agenda and org-agenda-list.

So, to sum up, I see three possible solutions:

1) org-agenda-goto and org-agenda-goto-mouse jump to a line with the
   timestamp (old org-agenda-goto behavior), org-agenda-switch-to jumps
   to the heading (current org-agenda-goto behavior).
   (I vote for this).

2) org-agenda-switch-to gets the org-agenda-goto behavior.
   (I can live with it, but I think it is less pretty.)

3) org-agenda-goto and *-mouse (or just their key bindings) do
   different things in plain agenda and in agenda-list.
   (Seems complicated.)

Cheers,
  Jan