Michael G

We discussed my previous post about optimizing the performance of a Rails application with SQL within the Belarus User Group. During the discussion, we also had some fun experimenting with ChatGPT, asking it to solve the same problem. The results were impressive. I shared the outcome in this post: https://blog.widefix.com/optimize-performance-of-rails-app-with-chatgpt/

The JRuby community is pleased to announce the release of JRuby 9.4.3.0.

• Homepage: https://www.jruby.org/
• Download: https://www.jruby.org/download

JRuby 9.4.x targets Ruby 3.1 compatibility.

Thank you to our contributors this release, you help keep JRuby moving forward! @byteit101, @donv, @k77ch7, @rcrews

Ruby Compatibility

• Multiple fixes for keyword arguments edge cases. #7676, #7755, #7804, #7816

Standard Library

• Default gem updates:
  • rubygems 3.3.26 #7831
  • bundler 2.3.26 #7831
  • cgi 0.3.6 #7831
  • date 3.3.3 (stub gem on JRuby) #7831
  • ffi 1.15.5 #7831
  • jruby-openssl 0.14.1
  • net-http 0.3.0 #7831
  • net-protocol 0.1.2 #7831
  • reline 0.3.5 #7831
  • ffi-binary-libfixposix 0.5.1.1 to support Apple M1 (used by subspawn for improved subprocess management)
  • time 0.2.2 #7831
  • timeout 0.3.2 #7831
  • uri 0.12.1 #7831
• Bundled gem updates: #7831
  • net-imap 0.2.3
• Many fixes for BigDecimal behavior by @k77ch7. #7747, #7758, #7766, #7767, #7768, #7778, #7779, #7780, #7781, #7813, #7827
• Rippper now passes all tests related to the syntax_tree gem. #7776

Performance

• Various small optimizations based on Rails and other frequent use cases. #7522, #7720, #7722, #7783, #7789

JVM Integration

• InvokeDynamic support is now being tested across five major test suites on Java 17 and passes all tests that non-indy execution passes. #7797

79 Github Issues resolved for 9.4.3.0

• #1422 – GzipReader#mtime gives the wrong time
• #6607 – Clean up javadoc for stricter Java 11
• #7452 – Default external to locale as in CRuby
• #7474 – Rails 7 support
• #7522 – case/when optimizations
• #7579 – GLIBC compatibility with RHEL/CentOS 7 and JRuby 9.4.0.0
• #7673 – Enable more traces and reduce costs of others
• #7676 – ruby2_keywords should not warn when using …
• #7711 – Constants set by autoload have wrong line number in JIT
• #7712 – Use real stack trace for autoload const location
• #7714 – Enumerator cleanup
• #7717 – sass-embedded gem extension install fails with rake require error on Windows with JRuby 9.4.2.0
• #7718 – Deprecation cleanup
• #7719 – Fixes #7717. sass-embedded gem extension install fails with jruby-com…
• #7720 – More indy call optz
• #7722 – Eliminate duplication for optimized case/when build
• #7723 – passing specs
• #7724 – Syslog tags
• #7725 – More specs already fixed
• #7726 – Ruby-based SecurityManager can recurse forever
• #7728 – [feat] make Ruby#setWarningsEnabled actually useful
• #7734 – Object with finalizer dumps wrong data leading to: load': marshal data too short
• #7735 – Dir.home is no longer compatible with CRuby after JRuby 9.3
• #7738 – Fix Dir.home called with nil user
• #7739 – Don’t try to marshal non-serializable variables
• #7740 – irb error on Mac M1
• #7741 – some line numbers on warnings are off by 1
• #7743 – [ripper] NPE during parsing the rspec documentation
• #7744 – The File#absolute_path? implementation in JRuby 9.4.x.x causes issues with asciidoctor-pdf
• #7745 – Pathname#absolute? stopped working properly for URIs
• #7746 – Flaky test running rack on JRuby
• #7747 – Ruby 3.1 : fix Bigdecimal
• #7751 – Move arity-checking into variable-arity method bodies
• #7753 – Fix #7743. NPE during parsing the rspec documentation
• #7755 – Fixes #7676. ruby2_keywords should not warn when using …
• #7756 – Regression go away
• #7757 – Migrate regressions into jruby:spec or remove if already covered
• #7758 – Fix BigDecimal#sub with Object and significant digits
• #7761 – Can’t match rspec errors in 9.4.2.0
• #7764 – Mysterious “unused literal ignored” warnings
• #7765 – GH-6607 Clean up javadoc for stricter Java 11
• #7766 – Fix rounding mode constants of BigDecimal to use the same mri’s values
• #7767 – Fix BigMath#exp and #log error message for edge cases
• #7768 – Fix BigDecimal(0, NAN and INFINITY) to return always same object
• #7770 – Fixes #7769. regexp with / before mbc should work.
• #7774 – ripper presents an extra on_var_field for opt parameters in methods
• #7775 – Fixes #7774. ripper presents an extra on_var_field for opt params
• #7776 – Ripper fixes found while running syntax_tree gems tests.
• #7777 – fix try_convert error message
• #7778 – Fix BigDecimal#to_s for edge case
• #7779 – Fix bigdecimal/util’s #to_d and BigDecimal.interpret_loosely to return correct value for loosely case
• #7780 – Fix several BigDecimal failures
• #7781 – Add BigDecimal#power(n, prec)
• #7783 – Rails profiling optimizations
• #7789 – Java call optimizations
• #7790 – jruby –help displays removed taint level command line option
• #7791 – Remove obsolete command line option -T[level]
• #7793 – Macbook M1 JRuby 9.4.2.0 Could not open library ‘fixposix’ ‘libfixposix.dylib’
• #7797 – Fix recent regressions on master
• #7800 – Update time and uri for recent ReDoS issues
• #7801 – Update time to 0.2.2 and use date stub gem 3.3.3
• #7802 – Update Subspawn to fix #7793
• #7804 – Incorrect ArgumentError: wrong number of arguments
• #7807 – Update joni to 2.2
• #7808 – Update to backport9 1.13 for module javadoc fix
• #7813 – Add BigDecimal#quo(value, digits)
• #7814 – Update jzlib and fix mtime
• #7816 – Fixes #7804. Incorrect ArgumentError: wrong number of arguments
• #7818 – Small documentation issues regarding java options
• #7820 – Accessing Encoding from java code is broken on new thread
• #7824 – Use top self for evalScriptlet, since no frame yet
• #7826 – Fix Rational#truncate with an invalid value
• #7827 – Fix BigDeciaml#to_f
• #7828 – Improve verbiage of how to use properties.
• #7830 – ArrayIndexOutOfBoundsException matching regex against emoji string
• #7831 – Update stdlib gems and compat to 3.1.4
• #7832 – Update JNR versions
• #7833 – Update IRB tests to 1.7.0 version
• #7834 – Return to IRB 1.4.2 for now

o Fix double free corner case in ipmiseld.

o Support ISO 8601 date inputs to date range options in ipmi-sel.

o Support Xilinx OEM FRU records.

o Fix corner case in libipmimonitoring, support sensor if units 

  is “RPM per minute”, but just return “RPM” as the units.

o Fix portability compilation on cygwin.

o Fix typo in fiid template field that could lead to packet

  interpretation errors.

https://ftp.gnu.org/gnu/freeipmi/freeipmi-1.6.11.tar.gz

For those of you a little confused about what a postcode is, it’s effectively the same as a US zip code; a way of distilling a postal address down to but a few characters. Hence why some rogue auto-translate function in Windows 11 is occasionally switching ‘zip’ to ‘postcode’ in the UK’s Windows menus. As a translator myself, this is easy enough to explain. Either we’re looking at a terrible machine translation that wasn’t properly vetted, or a translator/reviewer not getting enough context to properly translate this string. As translators, we often get the absolute bare minimum to work with when it comes to software – usually just the strings, and if we’re very, very, very lucky, we might get a screenshot, but that’s a rarity. It’s easy to look at this and think the translator is an idiot, but without any context, some isolated strings, often delivered in a random order, can be incredibly hard to translate in a way that makes any sense in the target context. It’s just another way the software industry gets away with bottom-of-the-barrel effort, something no other industry is allowed to do. A random package of disposable paper plates has to adhere to more standards, controls, and checks than consumer software has to do. Managers in the consumer software industry face virtually no consequences for shipping the absolute bare minimum in quality, and unlike in any other industry, shipping broken garbage that never gets fixed is the norm, rather than the exception. There’s no other product category in our lives where we would tolerate the amount of brokenness that’s common in software. And, of course, software translations are no exception. It’s an easy target for managers to outsource and automate to “save money”. This is what it leads to.