https://github.com/sbt/sbt/pull/4143 kept failing, and while I was debugging,
I noticed that the target that it was downloading scala-compiler was in
/tmp/... for the remote instance.
This moves the fakeboot to a full path, which should make the build
less susceptible to failure due to random network issues.
Fixes#3482 take 3
There are two bugs related REPL and JLine.
1. JLine getting disabled (up arrow not working) on the second run of `console` task.
2. Typed characters not showing up even on the `console` task.
The first issue was fixed by #4054 which added `t.init`. When I noticed the second problem, I fixed it in #4082 (adds `t.restore`) but undid the first fix. This attempts to fix both the issues.
Previously we'd get in the build logs:
[error] params cannot be negated, it enables other arguments
and lots of wawrnings.
Now we just get lots of warnings without the non-fatal error message.
The existing filter caused SourceModificationWatch.watch to ignore
deleted files because !file.exists implies !file.isFile. The intention
of the filter was to exclude directories that had a name ending in
".scala".
This version of Jline fixes three things for Emacs users:
- ANSI colors are now enabled for Emacs.
- Terminal echo is now disabled for Emacs.
- History is enabled for all dump terminals.
Fixes#3482 take 2
I thought I tested #4054 using a local build, but when I ran 1.1.3, `console` did not display anything that I typed.
Switching to `usingTerminal` which calls `terminal.restore` similar to what I had in 1.1.1 fixes `console`.
This hot path was discovered by retronym using FlameGraph.
This removes intermediate creation of Array.
`filesModifiedBytes` in particular was bad as it was going through all `*.class` files, each generating an Array. This replaces that with `fileModifiedHash` that accepts `MessageDigest`.
According to the flamegraph, evalCommon footprint reduced from 4.5% to 3.6%.
Using `time sbt reload reload reload exit`, the median reduced from 41.450s to 39.467s.