summaryrefslogtreecommitdiff
path: root/main/client
Commit message (Collapse)AuthorAgeFilesLines
* Added changes to the MillClientMain to make it only use workers runningMark Hammons2019-03-131-3/+28
| | | | the JVM the client was launched with.
* Reviewed exitCode handlingTobias Roeser2019-02-251-5/+12
| | | | Propagate server stop because of version change as exit code.
* collapse boilerplate folder structure within src/ folders (#505)Li Haoyi2018-12-129-0/+0
| | | | | | * collapse boilerplate folder structure within src/ folders * .
* Fix MILL_CLASSPATH for Windows (#354)Joseph K. Strauss2018-05-301-9/+20
| | | | | | | | | | | | | | | | | | | * Use comma as separator in MILL_CLASSPATH There is no need to use environment-specific separator, especially since - other variables are using commas anyway, and - it is not sent to any system-level command * Fix whitespace * Use MILL_CLASSPATH for Windows * Use vm options file for client on windows * Remove overzealous distinct * Clean up unnecessary ceremony
* Migrate `ProxyOutputStream` to the `main.client` module, add unit/fuzz tests ↵Li Haoyi2018-05-228-51/+193
| | | | to make sure it works
* optimize output streaming to allow batch writesLi Haoyi2018-05-221-24/+12
|
* Preserve caches between interactive and client/server mode (#342)Li Haoyi2018-05-191-14/+2
| | | | We were incorrectly duplicating the JDK classpath as part of the application classpath when we spawned the Mill server from the Mill client. This makes the transmission of application classpath to the Mill server explicit via an environment variable, so we don't end up including random things from the client classloader hierarchy that we didn't expect
* WIP keep mill server alive if you Ctrl-C during --watch (#339)Li Haoyi2018-05-193-45/+66
| | | | | | | | * wip * Clean up more resources in the Mill client after every command * catch and ignore SIGINT in Mill server to make it survive Ctrl-C on the client
* rename modules scalaworker -> scalalib.worker, client -> main.clientLi Haoyi2018-04-127-0/+528