Commit Graph

31 Commits

Author SHA1 Message Date
Eugene Yokota f77c15496b add JDK 9 support 2017-02-15 04:57:47 -05:00
Eugene Yokota acd610be67 Fix checkJava function 2017-01-16 03:09:29 -05:00
eugene yokota 86afaaebb4 Merge pull request #134 from Atry/patch-2
Don't set MaxMetaspaceSize on Java 8 by default
2017-01-11 10:29:38 -05:00
Eugene Yokota f14b165523 Fix java version detection
Previously awk was used to grab the full Java version such as 1.8.0_91.
While this is more accurate, 1.8.0_91 is not a number that can be compared by bash, and thus JDK8 detection logics were failing.
Fixes #135
2017-01-11 00:20:38 -05:00
杨博 (Yang Bo) e1d64533f3 Don't set MaxMetaspaceSize on Java 8 by default 2017-01-11 12:00:34 +08:00
Sam Halliday d9da67979a -java-home should set PATH for javac usage 2016-08-25 20:16:35 +01:00
Eugene Yokota 843a60734b Fixes #104. Allow individual memory options to be controlled
When a `-J` option is passed in such as `-J-XX:MaxPermSize=1001M`, it
will override only that option.
2016-02-19 17:46:05 -05:00
Mihai Capotă 0d9806b9ff Check memory options in JAVA_TOOL_OPTIONS
JAVA_TOOL_OPTIONS is used by the JVM [1] so it must also be checked for memory options that conflict with the ones set in get_mem_opts.

[1] https://docs.oracle.com/javase/8/docs/technotes/guides/troubleshoot/envvars002.html
2016-02-17 11:26:13 -08:00
Eugene Yokota 40167af167 Handle SBT_OPTS
Fixes sbt/sbt-launcher-package#99 formerly known as sbt/sbt#2194.
get_mem_opts() is a bash function that generates memory related
options. This change makes it return `""` the SBT_OPTS variable
contains memory-related options already.
2016-01-21 23:26:25 -05:00
William Waites 892de8e7d7 Set resource limit for ${java_cmd} -version in sbt initialisation
Java is a little anti-social and attempts to lazily allocate all
of system memory, even for simple operations such as printing out
the version. This causes sbt to fail to start in environments
where resources are limited (i.e. ulimit(1)). This setup is common
on shared infrastructure such as scientific computing clusters
where because of the resource limit not being specified sbt cannot
be used.

The limit is set to 512MB which ought to be ample and is in any case
the default from sbtconfig.txt. A better patch would use the limit
specified there but it isn't clear that that is worth the effort.
2015-01-20 13:16:33 +00:00
Kousuke Saruta 477f271d45 Fixed wrong debug option 2014-11-21 01:10:48 +09:00
Josh Suereth 09d906de4b Merge pull request #84 from KaiXinXiaoLei/master
Delete unnecessary function
2014-11-20 09:55:27 -05:00
KaiXinXiaoLei bb06511109 Delete unnecessary function 2014-11-13 20:33:20 +08:00
Kousuke Saruta be20baace0 Fixed debug option 2014-11-07 15:02:02 -08:00
Stephen Nancekivell 10e95ef240 use JAVA_OPTS instead of java_opts. 2014-11-02 12:52:56 +11:00
Josh Suereth 7c5bf6f5f6 Merge pull request #75 from raboof/remove_unused_variable_mergable
Remove unused variable script_dir
2014-10-30 11:03:05 -04:00
Roberto Tyley cb12ec1eaa Use MaxMetaspaceSize rather than MaxPermSize on Java 8
@jroper points out that class metadata has to go somewhere, and it's new
name is 'Metaspace' - not 'PermGen':

https://github.com/sbt/sbt-launcher-package/pull/66#issuecomment-57407643

http://java.dzone.com/articles/java-8-permgen-metaspace
2014-10-01 09:24:01 +01:00
Arnout Engelen bb21b9ce09 Remove unused variable script_dir
This variable used to be set based on either $HOME or $script_path. However,
$script_path is no longer set, so when $HOME is unset there will be an
'dirname: missing operand' error message. $script_dir is no longer used
either, so we can safely remove it entirely.
2014-09-11 17:01:54 +02:00
Roberto Tyley f716915f52 Don't pass MaxPermSize to Java 8+
This change stops this annoying message being written to stderr (which can
cause IntelliJ to freak):

```
Java HotSpot(TM) 64-Bit Server VM warning: ignoring option MaxPermSize=256m; support was removed in 8.0
```

See also:

https://bugs.openjdk.java.net/browse/JDK-6964458
https://github.com/sbt/sbt-native-packager/issues/203
https://github.com/typesafehub/activator/issues/422
2014-04-23 00:24:17 +01:00
Roberto Tyley c2deacbcbb Only calculate memory opts if we need them...
...don't bother if won't use them due to them already being defined.
2014-04-23 00:02:12 +01:00
Josh Suereth 8e4dbbaf0c -mem options only used if not otherwise specified directly. 2014-01-06 09:39:29 -05:00
Josh Suereth 1a07150560 Turns out the cygwin fix involved a bit more work.
We can't exec bash, because we can't give over our process to sbt
if we want to turn echo back on later.
2013-07-30 19:49:17 -04:00
Josh Suereth 19f9330c32 Fixes #53 - echo disabled after sbt exits.
Bad flag in bash.  Wish this thing had a compiler....
2013-07-30 19:38:33 -04:00
Josh Suereth 2b462315e5 Reduce necessary memory 2013-07-14 15:36:07 -04:00
U-jsuereth-win\jsuereth d50c597318 Initial fixes for cygwin issues #50 2013-07-14 15:27:26 -04:00
Josh Suereth 53f9de2735 Migrating to new native packaging plugin abstractions.
* Use new native packaging abstractions for windows mappings.
* Update universal sbt script to be cygwin friendly
* Modify sbt.bat for universal layout
* Fix some stty icanon (no)echo bugs in launcher script.
2013-06-29 10:09:42 -04:00
Josh Suereth a99f6fffbf Upgrade to in-progress sbt-native-packager plugin, including new support for universal->debian/rpm packages. 2013-06-27 16:43:31 -04:00
Josh Suereth 55e0dfdc65 Removed bad practices from sbt bash script.
* No set XYZ arguments allowed.  BAAD sbt practice.
* No longer swaps launcher.  New SBT launchers can handle old SBT.
* SBT script now uses sbt-launch-lib.bash.
2012-07-17 16:37:19 -04:00
Josh Suereth b8173fc41b Removed autodownload from the scripts. 2012-05-09 09:40:10 -04:00
Josh Suereth 70d0b31904 Bumped version numbers for 0.11.3 2012-05-07 08:19:44 -04:00
Josh Suereth ff05caba8b updated to latest native packager, added universal 'zip' generation to the build. 2012-03-15 14:22:10 -04:00