First ever bundle install, stack level too deep
bundle, bundler, ruby-on-rails, rubygems, stack
Solution
Cygwin's pthread_attr_getstack reports about 90k by default. I think Cygwin is reporting the committed (used-so-far) stack size, not the reserved (total available) size. Then Ruby is treating that (i.e. 90k) as the maximum stack size ever.
To change the initial committed stack size:
peflags -X409600 `which ruby`
Note the capitalized -X. Lowercase -x changes reserved stack size, not initially-committed stack space.
This fixes all Ruby+Rails+Cygwin issues for me.
I don't know Ruby or Cygwin internals well enough to know which is the right fix, but it's probably one of these.
- Cygwin reports reserved stack size from pthread_attr_getstack.
- Ruby adds a build step to run the peflags command above.
- Ruby determines the available stack space a different way.
Problem
I have created a brand new rails project using the command `rails new qbc --database=mysql`. It creates all the files perfectly fine, but at the bundle install it errors out ``` $ bundle install Fetching gem metadata from https://rubygems.org/........... Fetching gem metadata from https://rubygems.org/.. Unfortunately, a fatal error has occurred. Please see the Bundler troubleshooting documentation at http://bit.ly/bundler-issues. Thanks! /usr/bin/bundle:23: stack level too deep (SystemStackError) ``` `gist.github.com/3956513` I have searched and searched for the solution to this issue but I can't seem to find anyone else who has experienced it. I am developing on Cygwin and I wouldn't be surprised if that has something to do with it. I tried creating a Gemfile with just the source and one gem in it in an empty directory and bundle install still gives the same error. I have followed all of the troubleshooting steps, reinstalled Cygwin and all packages, everything. What keeps catching my eye is the `Fetching gem metadata` twice...is it possible that bundler is caught in some kind of loop?