Download Expectation Failure for SHA dependency

[Justin Blank]

I'm receiving the following error when running stack build, and I'm not sure whether it reflects something wrong with my configuration or stack, or Stackage. Can anyone give me guidance on where to look?

Download expectation failure: content hash (SHA512)

Expected: aa1178f08ad3364c504ee790dac9b0774b7461fa7ea0a8f2c3ad82943d109602debcf4a8a83090815fef487ceac40019f8b5594ab0043f36e019d69bc5f76853 (ByteString)

Actual:   91f514e1731ef3c2063ff5ab6156242755b2b2b462d4a1f40f150d96cf714f7b7b5b19c7705e5129bbb9876216ca86ff52e49cd7604d83e4b4150172c9bda228

For: https://s3.amazonaws.com:443/hackage.fpcomplete.com/package/SHA-1.6.4.2.tar.gz

I'm running windows 7, 32-bit, with 

Stack Version 0.1.10.1, Git revision 83ec40fc5fd6d300c73d23deb1f41d30168fc236 (2783 commits) i386

I created the project by running 

stack new downloadtest simple

and adding 

SHA

to build-depends. 

My cabal file:

name:                downloadtest

version:             0.1.0.0

synopsis:            Simple project template from stack

description:         Please see README.md

homepage:            http://github.com/githubuser/downloadtest#readme

license:             BSD3

license-file:        LICENSE

author:              Author name here

maintainer:          exa...@example.com

copyright:           2010 Author Here

category:            Web

build-type:          Simple

cabal-version:       >=1.10

executable downloadtest

  hs-source-dirs:      src

  main-is:             Main.hs

  default-language:    Haskell2010

  build-depends:       base >= 4.7 && < 5

                       , SHA

[Michael Snoyman]

This shouldn't be necessary, but can you try deleting the SHA tarball inside your stack directory? It should be at:

    %APPDATA%\stack\indices\Hackage\packages\SHA\1.6.4.2\SHA-1.6.4.2.tar.gz

If this fixes things, then we need to add some logic to Stack to automatically delete corrupted downloads (logic I thought was already present).

--
You received this message because you are subscribed to the Google Groups "haskell-stack" group.
To unsubscribe from this group and stop receiving emails from it, send an email to haskell-stac...@googlegroups.com.
To post to this group, send email to haskel...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/haskell-stack/d6e3720a-743c-4704-b0bd-f2d7ebd30c43%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[Dan Burton]

Download expectation failure: content hash

This message comes from the WrongDigest exception. It appears to me that this is being thrown because your download is legitimately (and repeatedly?) corrupted. Is there a proxy messing with your download or something?

Here's the code path for checking for corrupted downloads:

https://github.com/commercialhaskell/stack/blob/6ab16a8416f8aa2165a65b89ad6ae47a63ec842c/src/Network/HTTP/Download/Verified.hs#L235

getShouldDownload -- calls

fileMatchesExpectations -- calls (and catches WrongDigest :: VerifiedDownloadException)

hashChecksToZipSink -- calls

sinkCheckHash -- throws WrongDigest

I don't think the exception is being printed out from this code path, because on this code path the WrongDigest exception is caught rather than being allowed to propagate to the top.

(I've manually downloaded the file in question and can confirm that it has the expected sha512.)

Here's a test case that claims to demonstrate that an existing incorrect file will be replaced with the correct downloaded file:

https://github.com/commercialhaskell/stack/blob/6ab16a8416f8aa2165a65b89ad6ae47a63ec842c/src/test/Network/HTTP/Download/VerifiedSpec.hs#L118

-- Dan Burton

To view this discussion on the web visit https://groups.google.com/d/msgid/haskell-stack/CAKA2JgJceAYCpmJSQFm5DfrPBTqmACkhS%3Dr_3y5KBF_HvfC8Vw%40mail.gmail.com.

[Dan Burton]

On Tue, Dec 29, 2015 at 1:21 AM, Dan Burton <danburt...@gmail.com> wrote:

Here's the code path for checking for corrupted downloads:

What I meant to say was, "Here's the code path for checking for corrupted exiting files."

-- Dan Burton

[Dan Burton]

This actually sounds very similar to an issue Michael had a long time ago:

https://github.com/commercialhaskell/stack/issues/372

We had a hard time reproducing it then, and I presume the same might be true of this case as well. Maybe that means there really is a bug in there somewhere.

-- Dan Burton

[Justin Blank]

Thanks a lot--after deleting the file, a subsequent stack build worked. 

I should note that in my case, the file is under C:/stack_root, not the AppData folder.

I retained the SHA-1.6.4.2.tar.gz file for comparison, just in the off chance that it is helpful (but I assume a synthetic test case would be easy enough to produce). Let me know if there is any other information I can provide that would be relevant to a bug fix.

[Dan Burton]

A few questions if you have the time.

Did you check with an outside tool that the corrupt file had the "actual" sha512 as reported by stack?

Can you reproduce the problem by copying the corrupt file back to where it was?

To view this discussion on the web visit https://groups.google.com/d/msgid/haskell-stack/eb66be51-1269-4aea-90ca-64145ce07a79%40googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

--
-- Dan Burton

[Michael Snoyman]

I was just able to reproduce this problem by changing a few bytes inside the file, but keeping the file the same size. I'll see about a patch to fix that.

To view this discussion on the web visit https://groups.google.com/d/msgid/haskell-stack/CALSygwcHZQHKNq%3DsMoJSJyr1VZPngvHLSGHO_i%3Dd2piivWWsdQ%40mail.gmail.com.

[Michael Snoyman]

I think this is now resolved on master with:

https://github.com/commercialhaskell/stack/commit/54a40cebf4df39bc9b541f173173d96681c5426b

You can try `stack upgrade --git` to test out the change.