bug#2176: 2176 proposal

bug#2176: 2176 proposal

[Zachary Kanfer]

[-- Attachment #1: Type: text/plain, Size: 1854 bytes --]

This patch makes the line include a link to run 'recover-session .
It also rephrases the text. The text used to be "If an Emacs session crashed
recently, type Meta-x recover-session RET \nto recover the files you were
editing." The rephrased text is "If Emacs crashed recently, you may recover
your previous session."
The text no longer tells you how to run the command (i.e., with Meta-x
recover-session RET). This is in line with the other commands on the page
(e.g., Emacs Tutorial, Open a File).
-Zachary Kanfer


=== modified file 'lisp/startup.el'
*** lisp/startup.el    2011-07-14 12:57:06 +0000
--- lisp/startup.el    2011-07-25 04:36:03 +0000
*************** a face or button specification."
*** 1574,1586 ****
                     auto-save-list-file-prefix)))
      t)
         (fancy-splash-insert :face '(variable-pitch font-lock-comment-face)
!                 "\nIf an Emacs session crashed recently, "
!                 "type "
!                 :face '(fixed-pitch font-lock-comment-face)
!                 "Meta-x recover-session RET"
                  :face '(variable-pitch font-lock-comment-face)
!                 "\nto recover"
!                 " the files you were editing."))

    (when concise
      (fancy-splash-insert
--- 1574,1585 ----
                     auto-save-list-file-prefix)))
      t)
         (fancy-splash-insert :face '(variable-pitch font-lock-comment-face)
!                 "\nIf Emacs crashed recently, you may "
!                             :link `("recover your previous session"
!                                     ,(lambda (_button) (call-interactively
'recover-session))
!                                     "Recover files from a previous
session")
                  :face '(variable-pitch font-lock-comment-face)
!                             "."))

    (when concise
      (fancy-splash-insert

[-- Attachment #2: Type: text/html, Size: 2166 bytes --]

bug#2176: 2176 proposal

[Eli Zaretskii]

> From: Zachary Kanfer <zkanfer@gmail.com>
> Date: Mon, 25 Jul 2011 23:13:00 -0400
> 
> This patch makes the line include a link to run 'recover-session .
> It also rephrases the text. The text used to be "If an Emacs session crashed
> recently, type Meta-x recover-session RET \nto recover the files you were
> editing." The rephrased text is "If Emacs crashed recently, you may recover
> your previous session."
> The text no longer tells you how to run the command (i.e., with Meta-x
> recover-session RET). This is in line with the other commands on the page
> (e.g., Emacs Tutorial, Open a File).

Recovering a crashed session is much more important than visiting a
tutorial or opening a file.  I would understand and support a
suggestion to _add_ a clickable link, but why would you want to remove
the information about invoking recovery by hand?  A user whose session
crashed may be under pressure, and not immediately recognize the
significance of the blue underlined portion of the text.

bug#2176: 2176 proposal

[Zachary Kanfer]

[-- Attachment #1: Type: text/plain, Size: 2819 bytes --]

My goal was to shorten up the text, as it's a little verbose. I think that
people are, by and large, used to the UI convention of "blue underlined"
means a link, but I get your point. What do you think would be the
appropriate part of the text to link? "recover-session" is the action, so
linking that would make sense, but the constant text changing (i.e.,
variable-pitch for the beginning of the line, fixed-pitch for "M-x",
fixed-pitch link for "recover-session", then variable-pitch again for the
rest of the line) gives it a jarring effect, at least to me. Perhaps I'm
overanalyzing. What do you think of the effects of this patch?
Do we also want "Meta-x" instead of "M-x", to make it easier for such users,
even though the usage is different from almost everywhere else in Emacs?
 -Zachary Kanfer

=== modified file 'lisp/startup.el'
*** lisp/startup.el     2011-07-14 12:57:06 +0000
--- lisp/startup.el     2011-07-28 05:56:34 +0000
*************** a face or button specification."
*** 1574,1586 ****
                               auto-save-list-file-prefix)))
        t)
         (fancy-splash-insert :face '(variable-pitch font-lock-comment-face)
!                           "\nIf an Emacs session crashed recently, "
!                           "type "
!                           :face '(fixed-pitch font-lock-comment-face)
!                           "Meta-x recover-session RET"
                            :face '(variable-pitch font-lock-comment-face)
!                           "\nto recover"
!                           " the files you were editing."))

    (when concise
      (fancy-splash-insert
--- 1574,1587 ----
                               auto-save-list-file-prefix)))
        t)

On Tue, Jul 26, 2011 at 12:58 AM, Eli Zaretskii <eliz@gnu.org> wrote:

> > From: Zachary Kanfer <zkanfer@gmail.com>
> > Date: Mon, 25 Jul 2011 23:13:00 -0400
> >
> > This patch makes the line include a link to run 'recover-session .
> > It also rephrases the text. The text used to be "If an Emacs session
> crashed
> > recently, type Meta-x recover-session RET \nto recover the files you were
> > editing." The rephrased text is "If Emacs crashed recently, you may
> recover
> > your previous session."
> > The text no longer tells you how to run the command (i.e., with Meta-x
> > recover-session RET). This is in line with the other commands on the page
> > (e.g., Emacs Tutorial, Open a File).
>
> Recovering a crashed session is much more important than visiting a
> tutorial or opening a file.  I would understand and support a
> suggestion to _add_ a clickable link, but why would you want to remove
> the information about invoking recovery by hand?  A user whose session
> crashed may be under pressure, and not immediately recognize the
> significance of the blue underlined portion of the text.
>

[-- Attachment #2: Type: text/html, Size: 3475 bytes --]

bug#2176: 2176 proposal

[Lars Magne Ingebrigtsen]

Zachary Kanfer <zkanfer@gmail.com> writes:

> What do you think of the effects of this patch?

The patch included just seems to remove the entire warning?

> Do we also want "Meta-x" instead of "M-x", to make it easier for such users,
> even though the usage is different from almost everywhere else in Emacs?

I'd say `M-x' instead of `Meta-x'.

>  -Zachary Kanfer
>
> === modified file 'lisp/startup.el'
> *** lisp/startup.el     2011-07-14 12:57:06 +0000
> --- lisp/startup.el     2011-07-28 05:56:34 +0000
> *************** a face or button specification."
> *** 1574,1586 ****
>                                auto-save-list-file-prefix)))
>         t)
>          (fancy-splash-insert :face '(variable-pitch font-lock-comment-face)
> !                           "\nIf an Emacs session crashed recently, "
> !                           "type "
> !                           :face '(fixed-pitch font-lock-comment-face)
> !                           "Meta-x recover-session RET"
>                             :face '(variable-pitch font-lock-comment-face)
> !                           "\nto recover"
> !                           " the files you were editing."))
>
>     (when concise
>       (fancy-splash-insert
> --- 1574,1587 ----
>                                auto-save-list-file-prefix)))
>         t)
>

-- 
(domestic pets only, the antidote for overdose, milk.)
  bloggy blog http://lars.ingebrigtsen.no/

bug#2176: Observations on the splash screen - and signal-process

[Chong Yidong]

storm@cua.dk (Kim F. Storm) writes:

> 1) Shouldn't "Meta-x" be "M-x" here ?
> 2) Couldn't we make this a "clickable" "Recover Session" button
>    like the Open file... instead of the text?
> 3) Maybe rephrase it to something like:

Done in trunk, thanks.

> Maybe emacs should check the saved sessions and filter out the
> currently running sessions ...  
> A simple (signal-process PID 0) would suffice -- unfortunately, the
> doc string for signal-process does not describe the return value
> in this case (currently it is 0 if the process exists and -1 otherwise
> - but better return values would probably be t or nil).

I don't think this would be strictly correct, since pids are reusable.
Anyway, it seems like over-engineering.

> Finally, wouldn't it make sense if (signal-process PID nil) checked
> for the existense of process PID - using signal 0 is probably
> non-portable.

I don't think it makes sense to overload `signal-process' for this
purpose; there is already `process-attributes'.