add fromList tests #663
add fromList tests #663
Conversation
| -- | ||
| -- Uses `reallyUnsafePtrEquality#`. | ||
| isUnit :: () -> Bool | ||
| isUnit = same () |
There was a problem hiding this comment.
I'm wondering whether I'm reinventing a wheel here. This seems closely related to what ChasingBottoms is doing, but I didn't find anything like this there.
There was a problem hiding this comment.
This smells funny/fragile to me. I would generally feel more comfortable with an approach that uses unsafePerformIO to find out when/how many times things are evaluated.
There was a problem hiding this comment.
Oh, but of course we can't always do that .... Hrmmm. Could you at least rename same to match our internal pointer equality function?
There was a problem hiding this comment.
Oh I wasn't aware of ptrEq, or I would have used that; will change.
I should probably make the tests conditional on __GLASGOW_HASKELL__ being defined... and maybe make isUnit tristate? (Yes | No | Maybe; the last indicating that the result is unknown.)
There was a problem hiding this comment.
Oh I wasn't aware of
ptrEq, or I would have used that; will change.
Ah it's not so easy because the module is hidden, and we probably want to keep the library parts of containers-tests and containers the same.
I should probably make the tests conditional on
__GLASGOW_HASKELL__being defined... and maybe make isUnit tristate? (Yes | No | Maybe; the last indicating that the result is unknown.)
I will go with Maybe Bool.
There was a problem hiding this comment.
Given your later comment, I stuck with Bool. I did add some logic to deal with non-ghc implementations though.
|
Do you consider the test or the code correct? Or neither? |
|
I believe the test is correct (it will pass once #658 is merged). |
|
Can you rebase please?
…On Sun, Jul 14, 2019, 5:41 PM Bertram Felgenhauer ***@***.***> wrote:
I believe the test is correct (it will pass once #658
<#658> is merged).
—
You are receiving this because you commented.
Reply to this email directly, view it on GitHub
<#663?email_source=notifications&email_token=AAOOF7P3SMSG4PCCWRFTSNTP7OMPTA5CNFSM4H6667HKYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGODZ4N36A#issuecomment-511237624>,
or mute the thread
<https://github.com/notifications/unsubscribe-auth/AAOOF7ISDKWPARUR7FJ43H3P7OMPTANCNFSM4H6667HA>
.
|
Done. (I don't know whether the force-push itself causes a notification.) |
|
The test suite is already pretty GHC-specific, so I wouldn't worry about
that. Tristate ... not sure. Depends what you're trying to check. Add more
comments?
…On Sun, Jul 14, 2019, 9:46 PM Bertram Felgenhauer ***@***.***> wrote:
***@***.**** commented on this pull request.
------------------------------
In containers-tests/tests/Utils/IsUnit.hs
<#663 (comment)>:
> +{-# LANGUAGE MagicHash #-}
+
+module Utils.IsUnit (isUnit) where
+
+import GHC.Exts
+
+same :: a -> a -> Bool
+same x y = case reallyUnsafePtrEquality# x y of
+ 0# -> False
+ _ -> True
+
+-- | Check whether the argument is a fully evaluated unit `()`.
+--
+-- Uses `reallyUnsafePtrEquality#`.
+isUnit :: () -> Bool
+isUnit = same ()
Oh I wasn't aware of ptrEq, or I would have used that; will change.
I should probably make the tests conditional on *GLASGOW_HASKELL* being
defined... and maybe make isUnit tristate? (Yes | No | Maybe; the last
indicating that the result is unknown.)
—
You are receiving this because you commented.
Reply to this email directly, view it on GitHub
<#663?email_source=notifications&email_token=AAOOF7JEXKJCY4WTQVETO63P7PJIBA5CNFSM4H6667HKYY3PNVWWK3TUL52HS4DFWFIHK3DMKJSXC5LFON2FEZLWNFSXPKTDN5WW2ZLOORPWSZGOB6L6IRA#discussion_r303272752>,
or mute the thread
<https://github.com/notifications/unsubscribe-auth/AAOOF7MOBKPA54D5G7WIGOLP7PJIBANCNFSM4H6667HA>
.
|
|
Nothing new from my side... the branch rebased cleanly on current master, and the tests currently pass (for me). |
- the function is rather peculiar in that in the presence of duplicate keys, the first value is not evaluated, but all the others are evaluated. See also #473.
- we check most functions that create or update entries, namely singleton, insert, insertWith, fromList, fromListWith, fromAscList, fromDistinctAscList, and fromAscListWith
This is adding some test cases that were previously part of #658.
Note that one test is currently failing because
Data.IntMap.fromAscListis a bit lazier than expected.