[pacman-dev] [RFC] Proposed fix for FS#20538
https://bugs.archlinux.org/task/20538 The attached patch flushes the terminal input buffer shortly before prompting users for input to answer a question. On the one hand, I can see how this might be nice when you accidentally hit <enter> twice, or fat finger a key during a download and press <enter> at the prompt only to be oddly returned to your shell. On the other hand, I've occasionally abused this exact behavior to walk away from my computer in the middle of a large -Syu. Thoughts? Any issues with the code approach? dave
On Wed, Mar 23, 2011 at 10:41 AM, Dave Reisner <d@falconindy.com> wrote:
https://bugs.archlinux.org/task/20538
The attached patch flushes the terminal input buffer shortly before prompting users for input to answer a question. On the one hand, I can see how this might be nice when you accidentally hit <enter> twice, or fat finger a key during a download and press <enter> at the prompt only to be oddly returned to your shell. On the other hand, I've occasionally abused this exact behavior to walk away from my computer in the middle of a large -Syu.
Thoughts? Any issues with the code approach?
You only use it in util, mark it static and drop it from util.h. Is this stuff we should guard using autoconf? I think if you add it to AC_CHECK_HEADERS and AC_CHECK_FUNCS you'll have useful #defines usable from config.h. -Dan
On Thu, Mar 24, 2011 at 12:18:20AM -0500, Dan McGee wrote:
On Wed, Mar 23, 2011 at 10:41 AM, Dave Reisner <d@falconindy.com> wrote:
https://bugs.archlinux.org/task/20538
The attached patch flushes the terminal input buffer shortly before prompting users for input to answer a question. On the one hand, I can see how this might be nice when you accidentally hit <enter> twice, or fat finger a key during a download and press <enter> at the prompt only to be oddly returned to your shell. On the other hand, I've occasionally abused this exact behavior to walk away from my computer in the middle of a large -Syu.
Thoughts? Any issues with the code approach?
You only use it in util, mark it static and drop it from util.h.
Is this stuff we should guard using autoconf? I think if you add it to AC_CHECK_HEADERS and AC_CHECK_FUNCS you'll have useful #defines usable from config.h.
-Dan
Done, and done. I've made flush_term_input effectively a NOOP when tcflush isn't available. Changes are in my GH repo [1]. dave [1] http://goo.gl/qQG91
participants (2)
-
Dan McGee
-
Dave Reisner