You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Due to commercialhaskell/stack#2604 (comment), I took a look at the decodeTooLong and decodeTooShort tests. Looks to me like the issue is that the test assumes Int is 8 bytes, but on ARM it's likely only 4.
I also noticed that these two tests are out of date - they do not include poking the messageMagic. The test is expecting any PeekException, including "Wrong message magic, ...".
Due to commercialhaskell/stack#2604 (comment), I took a look at the
decodeTooLong
anddecodeTooShort
tests. Looks to me like the issue is that the test assumesInt
is 8 bytes, but on ARM it's likely only 4.I also noticed that these two tests are out of date - they do not include poking the
messageMagic
. The test is expecting anyPeekException
, including "Wrong message magic, ...".I've pushed a branch, https://github.com/fpco/store/tree/alternative-streaming-test-encode, with an alternative approach for generating the too-short message and too-long message. Surprisingly, neither causes an exception to be thrown! Is this a bug?
For now, disabling the tests as all they're testing is the magic number detection.
The text was updated successfully, but these errors were encountered: