pry-stack_explorer compatibility

[banister]

Hi,

as of tonight (25 jan) pry 0.9.8 has been released. I haven't tested thoroughly but it appears pry-nav still seems to work ok.

I have also officially released pry-stack_explorer and, pry-nav appears to work fine with it out of the box on ruby 1.9.2. In fact, they work together in almost exactly the way i would expect, with no changes to either one necessary.

However on Ruby 1.9.3 i encounter a segfault when issue a show-stack command after moving forward one line with next -- i am as yet unsure about the cause of this segfault but i will investigate (and hopefully fix) as soon as i get the chance.

If it's not too much to ask, i wonder if you could also investigate the compatibility between these two plugins, so that we can put a notice for users to indicate any problems.

Thanks in advance!

John

[nixme]

Awesome. I'll test it out and see if I can come up with a workaround for the segfault.

[MSch]

I encountered an error when simply stepping through a method via n. Not sure if it's related. The crash went away when I didn't load pry-stack_explorer.

https://gist.github.com/c15ae6ed49e612ab4e1c

[banister]

@MSch thanks, I noticed something similar too, thanks for the bug report. As a result, for the meantime, i do not think it's safe to use pry-nav at the same time as pry-stack_explorer. Have to choose one or the other currently i think ;)

We'll work on getting them working together soon though.

[nixme]

@banister, are there any open issues on binding_of_caller that might be causing the segfaults? set_trace_func is pretty finicky in the ruby interpreter though, but I just don't want to rule out weird interaction between the C extension and the tracer.

[banister]

@nixme None that i know of. The segfault relates to GC (when i looked at it in gdb), and there's been substantial changes in the way GC works in 1.9.3, so i think the issue lies in there. I will look into it soon as i get a chance.

How's the behaviour on 1.9.2 btw?

[banister]

@nixme segfault should be fixed in most recent version of binding_of_caller (v0.6.2). I think all that's left for compatibility is to remove the pry-nav internal frames from the navigable call-stack, then we should be good to go!

I'll release an update to stack_explorer that does this soon :) thanks again

[nixme]

Great, I'll add stack_explorer to my jazz_hands package (https://github.com/nixme/jazz_hands) once you do.

[banister]

@nixme hey i released a version of stack_explorer (0.3.3) that should work seamlessly with pry-nav, give it a spin, and let me know how you get on!

Try with the following example, remember to mix and match step / next with up and down and tell me if you get what you 'expect' behaviour to be:

require 'pry-nav'

def alpha
  x = "hello"
  binding.pry
  beta("good evening")
end

def beta(val)
  puts val
  y = "a secret message"
  gamma
  v = "strange days ahead"
end

def gamma
  message = "end of the line"
  puts message
  puts "bye!"
end

alpha

[banister]

@nixme I don't mean to bug you :) but have you had a go with this yet? I can never find anyone else to try it, so im still unsure if this issue should be closed