<headius>
funny that mri:stdlib is the flaky bit now
<headius>
it seems to migrate
<headius>
kares: don't know how far you got into investigating that Java 9+ jossl error "No password supplied for PKCS#11 store" but I see what triggers it
<headius>
it's hard to get a decent trace because it swallows the system error (we need to fix that and include cause) and you can't easily set OpenSSL.debug from the plugins
<headius>
now the weird part is that the error message comes from BC
<headius>
engineLoad in JDKPKCS12KeyStore
<headius>
somewhere between load and engineLoad has changed
<headius>
maybe in the handling of null password by JDK SPI for jsse
<headius>
hmm
<headius>
there's only a couple hops
<headius>
load goes to the SPI which goes to engineLoad
<headius>
the docs for the SPI even say you can pass null for password
<headius>
I did force the password to char[] {} once and it went to a different error
<headius>
something about the keystore not being open :-D
rdubya has quit [Ping timeout: 276 seconds]
jmalves has joined #jruby
shellac has joined #jruby
shellac has quit [Client Quit]
shellac has joined #jruby
rdubya has joined #jruby
fredmorcos has joined #jruby
<fredmorcos>
hello lopex
<fredmorcos>
so it seems WEBrick is better than puma "when n is small", I'm guessing as long as its request queue isn't full/blocking it works much better. but for a much larger number of requests (which potentially fill up the queue), puma is far ahead