Copies products to the workind directory, and the rest to the serviceTempDir of this service, both wrapped in SHA-1 hash of the file contents. This is intended to mimize the file copying and accumulation of the unused JAR file. Since working directory is wiped out when the background job ends, the product JAR is deleted too. Meanwhile, the rest of the dependencies are cached for the duration of this service.
Fixes#2460Fixes#2851
Ref #2707, #2708, #2469
Unlike the previous attempts at fixing the handling of build-level
keys, this change does not change the main parsing logic, which uses
`getKey` to retrieve the key from the key map.
The fact that shell worked pre-0.13.11 means that the parsing was ok.
What this changes is just the "example" keys supplied to the parser so
the tab completion works.
Fixes#2761
With sbt 0.13.13-RC1 rediscovered that the dependency pulled in from
Giter8 was affecting the plugins. To avoid this, this change splits up
the template resolver implementation to another module called
sbt-giter8-resolver, and it will be downloaded using Ivy into
`~/.sbt/0.13/templates/`, and then launched reflectively using Java as
the interface.
This adds `new` command, which helps create a new build definition. The
`new` command is extensible via a mechanism called the template
resolver,
which evaluates the arbitrary arguments passed to the command to find
and run a template.
As a reference implementation [Giter8][g8] is provided as follows:
sbt new eed3si9n/hello.g8
This will run eed3si9n/hello.g8 using Giter8.
[g8]: http://www.foundweekends.org/giter8/
The compiler interface subclasses `scala.tools.nsc.Global`,
and loading this new subclass before each `compile` task forces
HotSpot JIT to deoptimize larges swathes of compiled code. It's
a bit like SBT has rigged the dice to always descend the longest
ladder in a game of Snakes and Ladders.
The slowdown seems to be larger with Scala 2.12. There are a number
of variables at play, but I think the main factor here is that
we now rely on JIT to devirtualize calls to final methods in traits
whereas we used to emit static calls. JIT does a good job at this,
so long as classloading doesn't undo that good work.
This commit extends the existing `ClassLoaderCache` to encompass
the classloader that includes the compiler interface JAR. I've
resorted to adding a var to `AnalyzingCompiler` to inject the
dependency to get the cache to the spot I need it without binary
incompatible changes to the intervening method signatures.