Small string fixes for encoding compatibility issues #124
Conversation
We expect exceptions from catString, so just re-raise directly. This caused the ugly internal RuntimeException trace in ruby/rake#619. This does not fix that issue, but it properly lets the original Ruby exception propagate.
These builds compile the extension, so they need to also have the command line `javac` be Java 21 when building against JRuby 10.
headius
force-pushed
the
no_runtime_exception
branch
from
9ac74f4 to
f63359b
Compare
|
These fixes should be released ASAP once green. The bad null StringIO code will not segfault on JRuby, but it will produce an internal Java NullPointerException when checking modifiability (which we consider equivalent to a "crash"). |
|
I can do but how about you try it?
|
|
@kou I will give it a try and let you know when it's done. |
|
@kou Apparently I do not have push rights for the gem and |
|
Actually it looks like it tries to release but fails during the build: https://github.com/ruby/stringio/actions/runs/14066275506/job/39389435498#step:3:278 I'm not sure what this error is so I'll have to investigate separately. We need to get the -java gem out there now though. |
|
No problem. I'll fix it and release a new version. |
|
It worked when I re-run the job: https://github.com/ruby/stringio/actions/runs/14066275506/job/39408925962 |
|
BTW, let's run the CI jobs with pull request to detect problems before we release a new version: #127 |
Great idea! Thanks for your help! |
See ruby/rake#619. The first two fixes here are in service of fixing that failure, but are not the actual bug fix. See jruby/jruby#8711 for the actual fix.
The other two fixes get jruby-head running green in CI.
catStringexceptions as a Java RuntimeException. This caused the ugly output in Bump ruby/setup-ruby from 1.222.0 to 1.227.0 rake#619.writemust be converted first, since it may be negotiated to something other than its original encoding.setup-javatojruby-headtargets to ensure Java 21 is in PATH for JRuby 10.