v2.15.0

• Support for explicit examples that show up in help system, see #261

v2.14.0

• Add support for parsing multiple uses of a flag as an array of values (see #243 courtesy @JacobEvelyn)

v2.13.4

• Replace deprecated call to File.exists? with File.exist? (see #229 thanks @edusantana!)

v2.13.3

• When generating a config file, only use the primary/first name of flags and switches. Previously, your config would have duplicates for every alias of a flag or switch. See #218 (thanks @ZeroPointEnergy!)
• Fixes issue where boolean settings were not read from the config properly. Namely, false values didn't seem to work properly. See #217 (thanks @ZeroPointEnergy!)

v2.13.2

• Fix bug where stty exists, but returns 0,0 (notably in emacs). See #223. Thanks @wconrad!

v2.13.1

• Removed 1.9 Hash syntax in hopes this can work with 1.8 (see #214, thanks @annafw!)

Note that the CI build fails on 1.8 due to the tests using Faker, which requires i18n, which will not install on 1.8 any longer

v2.13.0

• Add ability to disable "auto-guessing" the command names. e.g. if you have a command named list but execute my_app li, GLI will guess that you meant list. This can be problematic in some contexts and now can be disabled by using autocomplete false. See #200 (thanks @chrisarcand)

v2.12.3

• Don't use [-f foo] for flags that are required (See #210 thanks @dividedmind)

v2.12.2

• Fix crashing bug in the _doc command.
• Dropped official support for 1.8.7. This version of Ruby is long-deprecated and while GLI should continue to work, the tools needed to develop GLI exhibit strange behavior that cause tests to fail. Rather than lock to older versions of the tools or waste time figuring out the problem, I've removed 1.8.7 from .travis.yml. I don't plan to introduce 1.9 or 2.0-only code to GLI, but am not going to spend time making sure it works on 1.8.7

v2.12.1

• Since use of arguments :strict requires that subcommand_option_handling :normal also be set, GLI will blow up if that's not the case (as opposed to just not working).
• Scaffolding now puts subcommand_option_handling :normal in the new app, as was originally intended, but not ever actually done for some reason o_O.
• Improved documentation around arguments