* [sup-devel] [PATCH] Bugfix: Don’t call Ncurses.getch when in shell�out mode
@ 2010-02-19 3:26 Michael Stapelberg
2010-02-27 8:23 ` [sup-devel] [PATCH] Bugfix: Don\xe2\x80\x99t call Ncurses.getch when in shell\x96out mode Rich Lane
0 siblings, 1 reply; 9+ messages in thread
From: Michael Stapelberg @ 2010-02-19 3:26 UTC (permalink / raw)
To: sup-devel
[-- Attachment #1: Type: text/plain, Size: 527 bytes --]
Hi,
(at least for me) the attached patch finally fixes the pinentry-ncurses
problems we were still having (as soon as you press a key, the screen is
garbage).
To quote the commit message:
Previously, when using threads, Ncurses.getch was called while
the gpg pinentry was running (as an example of using the shell_out
method). Now, the Ncurses mutex will be used to wait until shell_out
mode is finished.
Please have a look if this patch is working for you and merge it for the next
release.
Best regards,
Michael
[-- Attachment #2: 0001-Bugfix-Don-t-call-Ncurses.getch-when-in-shell_out-mo.patch --]
[-- Type: application/octet-stream, Size: 2207 bytes --]
From 4d660530beeb42d192e2da594621b81ff991df10 Mon Sep 17 00:00:00 2001
From: Michael Stapelberg <michael@stapelberg.de>
Date: Fri, 19 Feb 2010 04:20:49 +0100
Subject: [PATCH] =?UTF-8?q?Bugfix:=20Don=E2=80=99t=20call=20Ncurses.getch=20when=20in=20shell=5Fout=20mode?=
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit
Previously, when using threads, Ncurses.getch was called while
the gpg pinentry was running (as an example of using the shell_out
method). Now, the Ncurses mutex will be used to wait until shell_out
mode is finished.
---
bin/sup | 2 +-
lib/sup/buffer.rb | 11 +++++++++--
2 files changed, 10 insertions(+), 3 deletions(-)
diff --git a/bin/sup b/bin/sup
index 7824aca..c8eb5b9 100755
--- a/bin/sup
+++ b/bin/sup
@@ -243,7 +243,7 @@ begin
until Redwood::exceptions.nonempty? || $die
c = begin
- Ncurses.nonblocking_getch
+ Ncurses.nonblocking_getch BufferManager
rescue Interrupt
raise if BufferManager.ask_yes_or_no "Die ungracefully now?"
BufferManager.draw_screen
diff --git a/lib/sup/buffer.rb b/lib/sup/buffer.rb
index c82fca0..84b3d00 100644
--- a/lib/sup/buffer.rb
+++ b/lib/sup/buffer.rb
@@ -30,13 +30,19 @@ module Ncurses
def mutex; @mutex ||= Mutex.new; end
def sync &b; mutex.synchronize(&b); end
- def nonblocking_getch
+ def nonblocking_getch bm=nil
## INSANTIY
## it is NECESSARY to wrap Ncurses.getch in a select() otherwise all
## background threads will be BLOCKED. (except in very modern versions
## of libncurses-ruby. the current one on ubuntu seems to work well.)
if IO.select([$stdin], nil, nil, 0.5)
- c = Ncurses.getch
+ if bm && bm.shelled?
+ # If we got input while we are shelled, we will drop the input
+ # but use Ncurses.sync to wait until the shell_out is done.
+ Ncurses.sync { nil }
+ else
+ Ncurses.getch
+ end
end
end
@@ -216,6 +222,7 @@ EOS
def sigwinch_happened?; @sigwinch_mutex.synchronize { @sigwinch_happened } end
def buffers; @name_map.to_a; end
+ def shelled?; @shelled; end
def focus_on buf
return unless @buffers.member? buf
--
1.6.5
[-- Attachment #3: Type: text/plain, Size: 143 bytes --]
_______________________________________________
Sup-devel mailing list
Sup-devel@rubyforge.org
http://rubyforge.org/mailman/listinfo/sup-devel
^ permalink raw reply [flat|nested] 9+ messages in thread
* Re: [sup-devel] [PATCH] Bugfix: Don\xe2\x80\x99t call Ncurses.getch when in shell\x96out mode
2010-02-19 3:26 [sup-devel] [PATCH] Bugfix: Don’t call Ncurses.getch when in shell�out mode Michael Stapelberg
@ 2010-02-27 8:23 ` Rich Lane
2010-02-27 12:48 ` Michael Stapelberg
0 siblings, 1 reply; 9+ messages in thread
From: Rich Lane @ 2010-02-27 8:23 UTC (permalink / raw)
To: Michael Stapelberg; +Cc: sup-devel
Excerpts from Michael Stapelberg's message of 2010-02-18 22:26:37 -0500:
> Hi,
>
> (at least for me) the attached patch finally fixes the pinentry-ncurses
> problems we were still having (as soon as you press a key, the screen is
> garbage).
>
> To quote the commit message:
>
> Previously, when using threads, Ncurses.getch was called while
> the gpg pinentry was running (as an example of using the shell_out
> method). Now, the Ncurses mutex will be used to wait until shell_out
> mode is finished.
>
> Please have a look if this patch is working for you and merge it for the next
> release.
>
> Best regards,
> Michael
What's the reason for passing the BufferManager singleton as an
argument?
What does "drop the input" mean?
_______________________________________________
Sup-devel mailing list
Sup-devel@rubyforge.org
http://rubyforge.org/mailman/listinfo/sup-devel
^ permalink raw reply [flat|nested] 9+ messages in thread
* Re: [sup-devel] [PATCH] Bugfix: Don\xe2\x80\x99t call Ncurses.getch when in shell\x96out mode
2010-02-27 8:23 ` [sup-devel] [PATCH] Bugfix: Don\xe2\x80\x99t call Ncurses.getch when in shell\x96out mode Rich Lane
@ 2010-02-27 12:48 ` Michael Stapelberg
2010-02-27 18:12 ` Rich Lane
0 siblings, 1 reply; 9+ messages in thread
From: Michael Stapelberg @ 2010-02-27 12:48 UTC (permalink / raw)
To: Rich Lane; +Cc: sup-devel
Hi Rich,
Excerpts from Rich Lane's message of Sa Feb 27 09:23:39 +0100 2010:
> What's the reason for passing the BufferManager singleton as an
> argument?
I am passing the BufferManager because I could not figure out how to access
it from inside the Ncurses module. Maybe there is a better way, if so, please
tell me :-).
> What does "drop the input" mean?
It means returning nil instead of Ncurses.getch, so maybe a better way of
saying it is "We do not check for input because it would restart Ncurses
while in shell-out mode".
Best regards,
Michael
_______________________________________________
Sup-devel mailing list
Sup-devel@rubyforge.org
http://rubyforge.org/mailman/listinfo/sup-devel
^ permalink raw reply [flat|nested] 9+ messages in thread
* Re: [sup-devel] [PATCH] Bugfix: Don\xe2\x80\x99t call Ncurses.getch when in shell\x96out mode
2010-02-27 12:48 ` Michael Stapelberg
@ 2010-02-27 18:12 ` Rich Lane
2010-02-27 19:00 ` [sup-devel] [PATCH] Bugfix: Don't call Ncurses.getch when in shell-out mode William Morgan
2010-02-27 19:27 ` [sup-devel] [PATCH] Bugfix: Don\xe2\x80\x99t call Ncurses.getch when in shell\x96out mode Michael Stapelberg
0 siblings, 2 replies; 9+ messages in thread
From: Rich Lane @ 2010-02-27 18:12 UTC (permalink / raw)
To: Michael Stapelberg; +Cc: sup-devel
Excerpts from Michael Stapelberg's message of 2010-02-27 07:48:39 -0500:
> Hi Rich,
>
> Excerpts from Rich Lane's message of Sa Feb 27 09:23:39 +0100 2010:
> > What's the reason for passing the BufferManager singleton as an
> > argument?
> I am passing the BufferManager because I could not figure out how to access
> it from inside the Ncurses module. Maybe there is a better way, if so, please
> tell me :-).
The reason we use these ugly singleton *Manager objects is to avoid
exactly this kind of plumbing. Since Ncurses is outside the Redwood
namespace you'll need to refer to it as Redwood::BufferManager.
> > What does "drop the input" mean?
> It means returning nil instead of Ncurses.getch, so maybe a better way of
> saying it is "We do not check for input because it would restart Ncurses
> while in shell-out mode".
Ok, that's how I read the code. When I saw "drop" in the comment I
thought it was going to read input and throw it away.
_______________________________________________
Sup-devel mailing list
Sup-devel@rubyforge.org
http://rubyforge.org/mailman/listinfo/sup-devel
^ permalink raw reply [flat|nested] 9+ messages in thread
* Re: [sup-devel] [PATCH] Bugfix: Don't call Ncurses.getch when in shell-out mode
2010-02-27 18:12 ` Rich Lane
@ 2010-02-27 19:00 ` William Morgan
2010-02-27 19:03 ` Rich Lane
2010-02-27 19:27 ` [sup-devel] [PATCH] Bugfix: Don\xe2\x80\x99t call Ncurses.getch when in shell\x96out mode Michael Stapelberg
1 sibling, 1 reply; 9+ messages in thread
From: William Morgan @ 2010-02-27 19:00 UTC (permalink / raw)
To: sup-devel
Reformatted excerpts from Rich Lane's message of 2010-02-27:
> The reason we use these ugly singleton *Manager objects is to avoid
> exactly this kind of plumbing. Since Ncurses is outside the Redwood
> namespace you'll need to refer to it as Redwood::BufferManager.
I know it's fashionable to call singletons ugly nowadays, but I've never
quite bought into it. Classes are singletons too, and no one calls them
ugly...
(Well, I do, but for different reasons.)
--
William <wmorgan-sup@masanjin.net>
_______________________________________________
Sup-devel mailing list
Sup-devel@rubyforge.org
http://rubyforge.org/mailman/listinfo/sup-devel
^ permalink raw reply [flat|nested] 9+ messages in thread
* Re: [sup-devel] [PATCH] Bugfix: Don't call Ncurses.getch when in shell-out mode
2010-02-27 19:00 ` [sup-devel] [PATCH] Bugfix: Don't call Ncurses.getch when in shell-out mode William Morgan
@ 2010-02-27 19:03 ` Rich Lane
2010-02-27 19:29 ` William Morgan
0 siblings, 1 reply; 9+ messages in thread
From: Rich Lane @ 2010-02-27 19:03 UTC (permalink / raw)
To: William Morgan; +Cc: sup-devel
Excerpts from William Morgan's message of 2010-02-27 14:00:30 -0500:
> Reformatted excerpts from Rich Lane's message of 2010-02-27:
> > The reason we use these ugly singleton *Manager objects is to avoid
> > exactly this kind of plumbing. Since Ncurses is outside the Redwood
> > namespace you'll need to refer to it as Redwood::BufferManager.
>
> I know it's fashionable to call singletons ugly nowadays, but I've never
> quite bought into it. Classes are singletons too, and no one calls them
> ugly...
>
> (Well, I do, but for different reasons.)
You've never tried to write an Index testsuite :).
_______________________________________________
Sup-devel mailing list
Sup-devel@rubyforge.org
http://rubyforge.org/mailman/listinfo/sup-devel
^ permalink raw reply [flat|nested] 9+ messages in thread
* Re: [sup-devel] [PATCH] Bugfix: Don\xe2\x80\x99t call Ncurses.getch when in shell\x96out mode
2010-02-27 18:12 ` Rich Lane
2010-02-27 19:00 ` [sup-devel] [PATCH] Bugfix: Don't call Ncurses.getch when in shell-out mode William Morgan
@ 2010-02-27 19:27 ` Michael Stapelberg
2010-02-27 20:14 ` Rich Lane
1 sibling, 1 reply; 9+ messages in thread
From: Michael Stapelberg @ 2010-02-27 19:27 UTC (permalink / raw)
To: Rich Lane; +Cc: sup-devel
Hi Rich,
Excerpts from Rich Lane's message of Sa Feb 27 19:12:19 +0100 2010:
> The reason we use these ugly singleton *Manager objects is to avoid
> exactly this kind of plumbing. Since Ncurses is outside the Redwood
> namespace you'll need to refer to it as Redwood::BufferManager.
Alright. Should I change the patch and re-send it or can you change
these little details?
Best regards,
Michael
_______________________________________________
Sup-devel mailing list
Sup-devel@rubyforge.org
http://rubyforge.org/mailman/listinfo/sup-devel
^ permalink raw reply [flat|nested] 9+ messages in thread
* Re: [sup-devel] [PATCH] Bugfix: Don't call Ncurses.getch when in shell-out mode
2010-02-27 19:03 ` Rich Lane
@ 2010-02-27 19:29 ` William Morgan
0 siblings, 0 replies; 9+ messages in thread
From: William Morgan @ 2010-02-27 19:29 UTC (permalink / raw)
To: sup-devel
Reformatted excerpts from Rich Lane's message of 2010-02-27:
> You've never tried to write an Index testsuite :).
What is complicated about it?
--
William <wmorgan-sup@masanjin.net>
_______________________________________________
Sup-devel mailing list
Sup-devel@rubyforge.org
http://rubyforge.org/mailman/listinfo/sup-devel
^ permalink raw reply [flat|nested] 9+ messages in thread
* Re: [sup-devel] [PATCH] Bugfix: Don\xe2\x80\x99t call Ncurses.getch when in shell\x96out mode
2010-02-27 19:27 ` [sup-devel] [PATCH] Bugfix: Don\xe2\x80\x99t call Ncurses.getch when in shell\x96out mode Michael Stapelberg
@ 2010-02-27 20:14 ` Rich Lane
0 siblings, 0 replies; 9+ messages in thread
From: Rich Lane @ 2010-02-27 20:14 UTC (permalink / raw)
To: Michael Stapelberg; +Cc: sup-devel
Excerpts from Michael Stapelberg's message of 2010-02-27 14:27:45 -0500:
> Excerpts from Rich Lane's message of Sa Feb 27 19:12:19 +0100 2010:
> > The reason we use these ugly singleton *Manager objects is to avoid
> > exactly this kind of plumbing. Since Ncurses is outside the Redwood
> > namespace you'll need to refer to it as Redwood::BufferManager.
> Alright. Should I change the patch and re-send it or can you change
> these little details?
I'll handle it. Applied to master.
_______________________________________________
Sup-devel mailing list
Sup-devel@rubyforge.org
http://rubyforge.org/mailman/listinfo/sup-devel
^ permalink raw reply [flat|nested] 9+ messages in thread
end of thread, other threads:[~2010-02-27 20:14 UTC | newest]
Thread overview: 9+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2010-02-19 3:26 [sup-devel] [PATCH] Bugfix: Don’t call Ncurses.getch when in shell�out mode Michael Stapelberg
2010-02-27 8:23 ` [sup-devel] [PATCH] Bugfix: Don\xe2\x80\x99t call Ncurses.getch when in shell\x96out mode Rich Lane
2010-02-27 12:48 ` Michael Stapelberg
2010-02-27 18:12 ` Rich Lane
2010-02-27 19:00 ` [sup-devel] [PATCH] Bugfix: Don't call Ncurses.getch when in shell-out mode William Morgan
2010-02-27 19:03 ` Rich Lane
2010-02-27 19:29 ` William Morgan
2010-02-27 19:27 ` [sup-devel] [PATCH] Bugfix: Don\xe2\x80\x99t call Ncurses.getch when in shell\x96out mode Michael Stapelberg
2010-02-27 20:14 ` Rich Lane
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox