Datamapper DataMapper

Comments and changes to this ticket

• 

  Alex Coles June 14th, 2008 @ 01:35 PM

  • Title changed from “extlib specs hang when executed on JRuby” to “extlib pooling needs tweaking for JRuby”

  Previous hang resolved by adjusting pool_size (see commit 59b51a3

  However, the values I chose is pretty arbitrary. Its worth spending a little time here (possibly getting the JRuby folks to put in some time) to figure out the values needed to optimize for JRuby.

  As such, I changed the ticket title.

• 

  Dan Kubb (dkubb) November 28th, 2008 @ 02:35 AM

  • State changed from “new” to “open”
  • Assigned user changed from “Sam Smoot” to “Dan Kubb (dkubb)”

  Alex, can you try dkubb/extlib on github and see if the problem persists?

  dbussink made some changes that should've fixed this.

• 

  Dan Kubb (dkubb) November 30th, 2008 @ 01:43 PM

  • Tag changed from “extlib” to “concurrency, extlib, jruby, pool, pooling”
  • State changed from “open” to “hold”
• 

  Alex Coles November 30th, 2008 @ 01:53 PM

  I pulled down the latest code this evening (commit #1fe1889 )

  and I can confirm the hang issue is fixed. However, I am encountering two failures in JRuby 1.1.5 (and notin MRI), one of which is related to pooling:

  
  1)
  'Object#encoded_hash returns the encoded hash like the value in the default Object#inspect' FAILED
  expected: "#<Oi:0x1187f5b>",
       got: "#<Oi:0x230feb6>" (using ==)
  spec/object_spec.rb:127:
  
  2)
  'Extlib::Pooling should wake up the scavenger thread when exiting' FAILED
  expected false, got true
  spec/pooling_spec.rb:151:
  

• 

  Dirkjan Bussink November 30th, 2008 @ 02:12 PM

  • State changed from “hold” to “resolved”

  The second issue is a JRuby bug and will be fixed in 1.1.6, the first failure is because Object#encoded_hash depends on the MRI implementation. Both are actually not critical to properly running DM with JRuby.