Implement a readline alternative on Windows

[murarth]

The recommended method of using GNU Readline on Windows is to install Cygwin, which is an unappealing option for some users.
A Windows version of the rusti::readline module could be implemented using Windows' console API.

[murarth]

cc @retep998

[bombless]

Hmm I just copy 2 dll files of readline to rustlib directory to compile rusti on Windows.
What is that cygwin thing?

[murarth]

Oh. Well, I suppose it would still be nice to remove the unnecessary dependency. Most Unix systems have readline installed, but it's an extra step for most Windows users.

[bombless]

Hmm maybe we first remove the feature to scroll up line history for Windows, then the fix will be easier.
I cannot scroll up because the up arrow key function differently in MSYS/CYGWIN shell.
(I believe this is a bug long time ago but it seems that nobody cares.)

[bombless]

(like this easy bombless@980c575 )

[bombless]

nvm I'll try make another crate to wrap readline and then do something like this (which looks awful)

[murarth]

@bombless: I don't intend to remove any features for either Windows or Unix platforms. The goal is to continue using readline on Unix systems, but use an alternative on Windows that provides basic functionality. The idea being that Unix users will generally expect a readline-like interface, as is used by popular shell and REPL utilities.

I don't know what sort of interface popular Windows shells and REPLs have, but if the Windows alternative has line editing (which I'm to understand is part of Windows' console API) and history support, that would be a good start.

I don't think any crazy trickery with dependencies should be necessary. The Unix dependency on readline is inferred by the extern "C" definition in readline.rs. Equivalent Windows functionality may need a Windows-specific dependency, but that shouldn't be much trouble.

[akavel]

maybe https://github.com/kkawakam/rustyline ? I don't know how well it works, only that it's the first seemingly pure-Rust Google result for "rust readline windows" for me as of 2016.08

[murarth]

I'm working on a cross-platform readline alternative that will soon be released and that will solve this issue.

[akavel]

In the meantime, I've tried plugging in rustyline as mentioned above. My fork [diff] seems to build successfully with nightly-2016-08-01; but on cargo run, it shows no prompt, and then after pressing [Enter] I'm seeing an assertion failure:

C:\prog\rusti>cargo run -v
       Fresh encode_unicode v0.1.3
       Fresh lazy_static v0.2.1
       Fresh libc v0.2.15
       Fresh cfg-if v0.1.0
       Fresh regex-syntax v0.2.2
       Fresh unicode-width v0.1.3
       Fresh log v0.3.4
       Fresh getopts v0.2.14
       Fresh memchr v0.1.7
       Fresh winapi-build v0.1.1
       Fresh winapi v0.2.5
       Fresh aho-corasick v0.4.0
       Fresh regex v0.1.46
       Fresh env_logger v0.3.2
       Fresh kernel32-sys v0.2.1
       Fresh advapi32-sys v0.1.2
       Fresh rustyline v1.0.0
       Fresh rand v0.3.12
       Fresh tempfile v1.1.3
       Fresh rusti v0.0.1 (file:///C:/prog/rusti)
    Finished debug [unoptimized + debuginfo] target(s) in 0.17 secs
     Running `target\debug\rusti.exe`

Assertion failed!

Program: C:\prog\rusti\target\debug\rusti.exe
File: C:/bot/slave/nightly-dist-rustc-win-gnu-64/build/src/llvm/lib/ExecutionEngine/MCJIT/MCJIT.cpp, Line 198

Expression: M->getDataLayout() == getDataLayout() && "DataLayout Mismatch"

This application has requested the Runtime to terminate it in an unusual way.
Please contact the application's support team for more information.
error: Process didn't exit successfully: `target\debug\rusti.exe` (exit code: 3)

C:\prog\rusti>rustc --version
rustc 1.12.0-nightly (7333c4ac2 2016-07-31)

Do you have any idea what can this be about? (At first sight, it doesn't seem to be broken rustyline, as running: C:\prog\rustyline> cargo run --example example — seems to work ok for me.)

[edit: please note I'm a total noob in Rust]

[akavel]

One more session in case you were interested, showing some slightly more interesting behavior before ultimately crashing too:

C:\prog\rusti>cargo run
    Finished debug [unoptimized + debuginfo] target(s) in 0.0 secs
     Running `target\debug\rusti.exe`
print
<input>:12:20: 12:25 error: unresolved name `print`. Did you mean the macro `print!`? [E0425]
<input>:12 println!("{:?}", { print });
                              ^~~~~
<std macros>:2:27: 2:58 note: in this expansion of format_args!
<std macros>:3:1: 3:54 note: in this expansion of print! (defined in <std macros>)
<input>:12:1: 12:29 note: in this expansion of println! (defined in <std macros>)
error: aborting due to previous error
print
<input>:12:20: 12:25 error: unresolved name `print`. Did you mean the macro `print!`? [E0425]
<input>:12 println!("{:?}", { print });
                              ^~~~~
<std macros>:2:27: 2:58 note: in this expansion of format_args!
<std macros>:3:1: 3:54 note: in this expansion of print! (defined in <std macros>)
<input>:12:1: 12:29 note: in this expansion of println! (defined in <std macros>)
error: aborting due to previous error
println!("Hello");
Assertion failed!

Program: C:\prog\rusti\target\debug\rusti.exe
File: C:/bot/slave/nightly-dist-rustc-win-gnu-64/build/src/llvm/lib/ExecutionEngine/MCJIT/MCJIT.cpp, Line 198

Expression: M->getDataLayout() == getDataLayout() && "DataLayout Mismatch"

This application has requested the Runtime to terminate it in an unusual way.
Please contact the application's support team for more information.
error: Process didn't exit successfully: `target\debug\rusti.exe` (exit code: 3)

[murarth]

Closing this because the new input reader, linefeed, is both cross-platform and not readline.

@akavel: If you're still experiencing that error on the new HEAD of master, go ahead and file an issue for it.