Update to run when connected to a pipe #147
Conversation
| @@ -4740,6 +4748,7 @@ def _rl_internal_char_cleanup() | |||
| def readline_internal_charloop() | |||
| lastc = -1 | |||
| eof_found = false | |||
| @is_pipe = (!@rl_outstream.winsize rescue true) | |||
There was a problem hiding this comment.
Do you think it could work if we checked !@rl_outstream.tty?, that way we don't introduce the io/console dependency...
There was a problem hiding this comment.
Do you think it could work if we checked !@rl_outstream.tty?
I checked every reasonably obvious means of testing for it in core, and I'm sure I tried tty. Nothing worked.
But, I should check in _rl_get_screen_size, as it might show the same info.
we don't introduce the io/console dependency...
I'm not bothered about a std-lib/default gem dependency, given that rb-readline isn't turning away collaborators. If this repo can make use of c code wrapped there, less support required here.
|
I'd love for us to get rb-readline working with pipes! As I'm sure you're by now aware, rb-readline can be a rather unforgiving mistress (due to the built up years of development on GNU Readline, of which this is basically a line for line port). Do you have any test programs to hand that you're using to verify it's doing what you expect? I'd like us to remove the dependency on |
|
I haven't taken the time to look at it, but ruby/reline might be something to look at. I suspect it's much more UTF-8 friendly... |
|
Closing due to availability of reline |
New work being done on Byebug showed that using rb-readline with pipes for input & output odes not function, as windows code defaults to keyboard API's.
This is not uncommon with testing, and it is one of the reasons rb-readline does poorly with ruby's readline tests. More fixes for that later.
This uses io/console's winsize attribute to determine this, and changes behavior accordingly.