Unify way of printing exceptions from within fibers

[Sija]

There was slight inconsistency how these exceptions were printed to STDERR (puts vs print), now they should be even. I've also added STDERR.flush missing from AtExitHandlers.run.

For the reference:

crystal/src/kernel.cr

Lines 404 to 405 in 5e059ab

	STDERR.print "Unhandled exception: "
	ex.inspect_with_backtrace(STDERR)

vs

crystal/src/fiber.cr

Lines 122 to 128 in 5e059ab

	if name = @name
	STDERR.puts "Unhandled exception in spawn(name: #{name}):"
	else
	STDERR.puts "Unhandled exception in spawn:"
	end
	ex.inspect_with_backtrace STDERR
	STDERR.flush

[Sija]

Weird, CI flunked (but just once) with the error below:

./bin/crystal build  -o .build/std_spec spec/std_spec.cr
.build/std_spec 

...............................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................*.................................................................................................................................................................................................................................................................................................................................................................................................................................................................................

Nil assertion failed (Exception)
  from src/class.cr:148:0 in 'not_nil!'
  from spec/std/signal_spec.cr:0:23 in '->'
  from src/signal.cr:255:3 in '->'
  from src/signal.cr:255:3 in 'process'
  from src/signal.cr:174:9 in '->'
  from src/fiber.cr:255:3 in 'run'
  from src/fiber.cr:29:34 in '->'
  from ???

FATAL: uncaught exception while processing handler for CHLD, exiting

Makefile:89: recipe for target 'std_spec' failed
make: *** [std_spec] Error 1

[straight-shoota]

This CI error happens infrequently. Don't know why. It's not related to this PR at least.

[straight-shoota]

Maybe puts everywhere would be better?

[Sija]

@straight-shoota Out of these two I prefer print, since puts will add a newline before the Exception message which ain't look good (to my eyes at least).

[Sija]

This CI error happens infrequently. Don't know why. It's not related to this PR at least.

I believe it's related to one of these:

crystal/spec/std/signal_spec.cr

Lines 5 to 7 in 5e059ab

	typeof(Signal::PIPE.reset)
	typeof(Signal::PIPE.ignore)
	typeof(Signal::PIPE.trap { 1 })

[straight-shoota]

Without a line break, this can become a pretty long line (consider a long fiber name and error message) and more difficult to read.

I'd prefer to add a line break after the "Unhandled exception" part.

[Sija]

I disagree, lines are longer by the size of "Unhandled exception: " string (which IMO is not much) and long names are seldom assigned to fibers, if at all (there are none in most of the cases).

Unhandled exception: Index out of bounds (IndexError)
  from /usr/local/Cellar/crystal/0.26.0/src/indexable.cr:596:8 in 'at'
  from /eval:1:1 in '__crystal_main'
  from /usr/local/Cellar/crystal/0.26.0/src/crystal/main.cr:104:5 in 'main_user_code'
  from /usr/local/Cellar/crystal/0.26.0/src/crystal/main.cr:93:7 in 'main'
  from /usr/local/Cellar/crystal/0.26.0/src/crystal/main.cr:133:3 in 'main'

vs

Unhandled exception:
Index out of bounds (IndexError)
  from /usr/local/Cellar/crystal/0.26.0/src/indexable.cr:596:8 in 'at'
  from /eval:1:1 in '__crystal_main'
  from /usr/local/Cellar/crystal/0.26.0/src/crystal/main.cr:104:5 in 'main_user_code'
  from /usr/local/Cellar/crystal/0.26.0/src/crystal/main.cr:93:7 in 'main'
  from /usr/local/Cellar/crystal/0.26.0/src/crystal/main.cr:133:3 in 'main'

Option no. 1 is more readable to me. Also, I'd prefer not to change already established format, just because of how an edge-case might affect it (spawn with loooong name).

[Sija]

Could this be reviewed and possibly included in the upcoming 0.26.1 release? It's a small change so it shoudn't take much time to get through it. @RX14 @bcardiff @ysbaddaden @asterite ?

[Sija]

Anyone care for 2nd review? 🏓 @asterite @sdogruyol