#__send__ wrong number of arguments N+1 for N (compile.fastest=true + JDK 7) #416
Comments
|
I've run into this issue too on jruby-1.7.2-dev. |
|
I seem to be getting this and do not have |
|
Sounds like a compiler failure. Is there a small (as small as a Rails app can be) test case? Also, can we confirm that this is not an application server dependent issue? (Though it is highly unlikely.) |
|
I have this issue as well. |
|
I have seen the issue with jruby-1.7.4 and with these jruby settings (non-default ones given). |
|
I feel like I committed a fix for the original issue, but I don't remember right now. There was a similar issue someone reported and the problem was not properly calling an aliased method name. @pmdb I would recommend not running with the fastsend setting. I'm curious if anyone has seen this issue in a recent JRuby without fastsend enabled. |
|
Thanks for the quick feedback. I also had this for compile.fastest enabled. If I enable use of invokedynamic, I get other problems. If there is anything I can do to help analyse this, let me know. My jrubyrc settings are currently very conservative .. only the threadpool seems to be harmless. compat.version=1.9 #compile.invokedynamic=true #compile.fastest=true #compile.fastsend=true thread.pool.enabled=true From: Charles Oliver Nutter [mailto:notifications@github.com] I feel like I committed a fix for the original issue, but I don't remember right now. There was a similar issue someone reported and the problem was not properly calling an aliased method name. @pmdb https://github.com/pmdb I would recommend not running with the fastsend setting. I'm curious if anyone has seen this issue in a recent JRuby without fastsend enabled. — |
|
I am interested in hearing more about the invokedynamic issues. send there should also be treated like a plain old call. |
|
The application is a RoR-3.2 application, we use a gem „active scaffold“ which produces much of the typical CRUD pages in a nice way. The server is mizuno / Jetty. The JDK for all tests was the latest Oracle JDK-7 (1.7.0-21). The jruby versions tested were 1.7.4 and a recent build of 1.7.5 (http://ci.jruby.org/view/JRuby%20Core/job/jruby-dist-master/1266/artifact/dist/) The results were the same for both jruby versions, the attached traces are from the test with 1.7.4 Test-1 compat.version=1.9 thread.pool.enabled=true è application works as expected Test-2 compat.version=1.9 thread.pool.enabled=true è application crashes quickly, see attached trace-test-2.txt è [ this trace was in the application log, not from stdout, so maybe there’s some other bug here and jruby is not the culprit ...] Test-3 compat.version=1.9 thread.pool.enabled=true è application crashes (not quite as quickly as above, but within 1 minute), see trace-text-3.txt Test-4 compat.version=1.9 thread.pool.enabled=true è application crashes (again about within 1 minute), see trace-text-4.txt From: Charles Oliver Nutter [mailto:notifications@github.com] I am interested in hearing more about the invokedynamic issues. send there should also be treated like a plain old call. — Exception: wrong number of arguments (1 for 0) Exception: wrong number of arguments (1 for 0) |
|
I can confirm this still happens with RoR 3.2. Does anyone have a workaround? |
|
1.7.x is EOL and this will not get fixed. Marking WONTFIX |
I am trying out my Rails app with jruby-1.7.0
After a few concurrent requests (under puma) I start getting the following error
Line 289 from connection_pool.rb:
Line 139 from monitor.rb:
Update: A few more runs show that:
__send__starts failingdate.rb:1786(5 for 4)compile.fastest=truein .jrubyrcThe text was updated successfully, but these errors were encountered: