Failing tests for the identifier parser #815
Conversation
| @@ -86,6 +86,9 @@ spec = do | |||
| it "parses identifiers enclosed within backticks" $ do | |||
| "`foo`" `shouldParseTo` DocIdentifier "foo" | |||
|
|
|||
| it "doesn't parse a composed function as an identifier" $ do | |||
| "'foo.bar'" `shouldParseTo` "'foo.bar'" | |||
There was a problem hiding this comment.
This looks potentially problematic: qualified identifiers should still be identified as DocIdentifier, no?
There was a problem hiding this comment.
But this is not a valid and qualified haskell identifier, right?
| "``infix``" `shouldParseTo` "`" <> DocIdentifier "infix" <> "`" | ||
|
|
||
| it "can parse identifiers in infix notation enclosed within single quotes" $ do | ||
| "'`infix`'" `shouldParseTo` "`" <> DocIdentifier "infix" <> "`" |
There was a problem hiding this comment.
I think this is the most problematic of all of these test failures, as exactly this syntax is described http://haskell-haddock.readthedocs.io/en/latest/markup.html#hyperlinked-identifiers.
Here's the test output:
expected: DocAppend (DocString "`") (DocAppend (DocIdentifier "infix") (DocString "`"))
but got: DocAppend (DocString "'") (DocAppend (DocIdentifier "infix") (DocString "'"))
There was a problem hiding this comment.
I just saw that this has been reported in #780.
|
I should add a test for #848 here. |
|
@sjakobi It would be nice to rebase this onto the 8.8 branch. I'm pretty sure almost all of these would now pass (except perhaps If you want to wait until after 8.8 is released, that's OK too (building all the deps is a pain...). |
|
Hi, thank you for this PR, but Haddock now lives full-time in the GHC repository! Let me know if you feel it is still needed, and I'll migrate it. :) |
While extending the
identifier parseridentifier parser tests I found the following test cases that I thought should pass.If someone can confirm that they should pass, I'll try to fix the parser.