Vaibhav Sagar

A potential downside to this approach is that projects with an inaccurate `setup.cfg` such as [pyrax](https://github.com/rackspace/pyrax/blob/master/setup.cfg) would incorrectly be seen as supporting Python 3. I think having Warehouse handle it...

Reproduced. I think this is another `pipes-zlib` or `pipes` bug: https://github.com/k0001/pipes-zlib/issues/24

No random blog posts either though.

I've implemented logic that separates a packfile into offset packfile entries, but there is no pre-existing logic for turning this into an index so I still need to write that.

I've written this function and it seems to work, but there are no tests for it yet.

Building the image offline is vastly preferable to using Packer, which does connect to an EC2 instance to create an AMI. This can be quite slow and requires an internet...

My first stab at upgrading `haskell-src-exts` is [here](https://github.com/lspitzner/exference/compare/master...vaibhavsagar:upgrade-haskell-src-exts), is it along the right lines?

How does `ghci` handle this? Edit: I just tried it and it seems to do the right thing: ``` Prelude> :set -Woverlapping-patterns Prelude> :{ Prelude| let f x = True...

Ah, it looks like the warning is displayed in the console, but not in the notebook itself. This is a similar issue to https://github.com/gibiansky/IHaskell/issues/1148.

You might need these two lines before running the tests: https://github.com/IHaskell/IHaskell/blob/e0c1a8d3fd0589ea41e23d58d8ebc5a05480eb81/release.nix#L27-L28 This is because the IHaskell tests import and use the `ihaskell` library within an interpreter session. Let me know...