Trailing newlines are calculated and represented in differ output. #290
Conversation
*Backfills hunk_generator with extra newline to account for 'true'
trailing newlines that should be included in diff calculation.
Example:
"cat\n".split("\n") => []
"cat\n\n".split("\n") => ["cat"]
"cat\n".split("\n", -1) => ["cat", ""]
"cat\n\n".split("\n", -1) => ["cat", "", ""]
*Removes extra newline representation from hunk_generator when crafting differ
output message.
|
Thanks for working on this! |
|
Looks like we have a failing spec, looking into it! |
|
Actually, @JonRowe any idea on what this could be https://travis-ci.org/rspec/rspec-support/jobs/142915946? I saw this encoding spec fail on master for me, and thought it might be something local. |
| def final_line(last_hunk) | ||
| output_message = last_hunk.diff(format_type).to_s.strip | ||
| # Checks last char of string for + or -, appends message on match | ||
| output_message.gsub(/(?<=[+-]$)\z/, "\\ No newline at end of input") |
There was a problem hiding this comment.
Rather than doing this, have you tried leaving the old logic as it was, and just checking for the missing trailing \n?
There was a problem hiding this comment.
I'm up for giving it another go! By 'old logic' do you mean
def expected_lines
@expected.to_s.split("\n", -1).map! { |e| e.chomp }
end
There was a problem hiding this comment.
The approach I think I'm suggesting, is to refactor expected_lines and actual_lines to check for a missing \n and add the text to the array if so, then allowing the differ to proceed as normal
There was a problem hiding this comment.
Something along the lines of:
def expected_lines
split_into_lines(@expected)
end
def actual_lines
split_into_lines(@actual)
end
private
def split_into_lines(string)
string << "\ No newline at end of input" if string[-1] != "\n"
string.split("\n").map! { |e| e.chomp }
endThis is psuedo code though, I haven't tried it directly
There was a problem hiding this comment.
Ah, okay.
Diff:
@@ -1,4 +1,5 @@
Chewy
+
bagel.
-\ No newline at end of input
+
Would this (above) be an acceptable output for
expected = "Chewy\nbagel."
actual = "Chewy\n\nbagel.\n"
The feature spec currently shows
Diff:
@@ -1,3 +1,5 @@
Chewy
+
bagel.
+\ No newline at end of input
There was a problem hiding this comment.
I'd like to not have the trailing + unless it actually represents a proper \n\n, if you can't get quite that, feel free to push it up anyway I'll take another look :)
There was a problem hiding this comment.
Eek, I've been away, but casually fiddling. Hope to move the discussion further, soon. Apologies!
There was a problem hiding this comment.
- If we are going to add text to indicate that there is an addition/deletion of an eof newline (instead of nothing at all), do we want this to happen before the actual or encoded string becomes and EncodedString?
- How do we want to handle internationalization? Would the message always be in English?
|
It seems like we are uncovering additional specific requirements of this issue. I’m glad to have moved the conversation forward, but I’m not comfortable with the solution being put forth given the invasiveness required. Thanks @CoralineAda for your support. Thanks @JonRowe for taking me along this far, I’ve learned a ton about open source, diffing, and RSpec! |
This is pretty much what @phiggins wrote in #71 :) but is up-to-date with master and rspec-expectations.
See commit message in 3a0af72 and rspec-expectations PR #290
@CoralineAda gave me the courage to dive in, and @alindeman let me borrow his eyes over coffee. I'm open to feedback, continued help, and working on next steps whatever they may be!
Fixes #70