Re: Simple lisp-tests.el and commit privs

[Barry OReilly]

> Oh, you're referring to known indentation problems, where you want
> to have the test but have it be marked as "expected".

Not just that. Suppose ruby.rb contains:

  if a == 2
    puts "hello"
    else
  puts "there"
  end

so as we can test general indentation which changes the buffer.

Expected success (current behavior):

--- ruby.rb     2013-08-20 11:29:39.000000000 -0400
+++ ruby.rb.new 2013-08-20 11:29:42.000000000 -0400
@@ -35,8 +35,8 @@
 
   if a == 2
     puts "hello"
-    else
-  puts "there"
+  else
+    puts "there"
   end
 
   if a == 2 then

Unexpected failure (if "else" didn't move -- is not current behavior):

   if a == 2
     puts "hello"
     else
-  puts "there"
+    puts "there"
   end
 
   if a == 2 then

Expected failure (current behavior)

@@ -52,11 +52,11 @@
   case a
   when "a"
     6
-  # Support for this syntax was removed in Ruby 1.9, so we
-  # probably don't need to handle it either.
-  # when "b" :
-  #   7
-  # when "c" : 2
+    # Support for this syntax was removed in Ruby 1.9, so we
+    # probably don't need to handle it either.
+    # when "b" :
+    #   7
+    # when "c" : 2
   when "d"  then 4
   else 5
   end

ERT has a means of distinguishing these. However, if we lump all these
in one test/indent/ruby.rb file and do as 'make ruby.rb.test' from one
ERT test, then these aren't distinguishable (in a simple way).

> Why not have a ert-with-temp-buffer which if the test fails (within
> the form), a file would be saved off with the buffer contents.
> Ideally the test case name would be in the generated filename.

Is this inadequate to address your debugging needs?