safe-exceptions VS rio-orphans

In safe-exception and uniftio it was decided to rethrow the original exception exactly because they decided to use uninterruptibleMask, see here for details.

unliftio (and safe-exceptions) contains a very controversial choice of of using uninterruptibleMask inside its bracket. The argument for it seem to come from this issue and comes from the fact that one of the most popular resource finalizers hClose is interruptible. This is a simplification. It is interruptible only if a file handle is used concurrently. Such usage of file handles is rather odd, and it suggest wrong architecture, for example leaking file handles using concurrency. When using file handles in synchronous setting, what withFile pattern encourages, hClose will not block and thus mask is enough.

https://github.com/commercialhaskell/rio#language-extensions which is cited as an example in simplehaskell's page on recommendations.

> Can we move to a better standard lib? Here Snoyman has put forward a great effort by releasing his classy-prelude, but iirc he also stopped using it.

He mentioned https://github.com/commercialhaskell/rio in the 1st article, it's interesting, I wasn't aware of it. (I am using classy-prelude but I might try it out.)

rio

It's worth mentioning that Snoyman's "Boring Haskell" is actually a fairly moderate position: if you look at his suggested list of language extensions, it's pretty broad (and fairly reasonable in my view).

via ByteString was recommended by Snoyman in the post Beware of readFile (referenced in the Haskell: The Bad Parts, part 1 too). But this has got a disadvantage compared to Data.Text.IO combined with hSetEncoding. This might be a good time to update Beware of readFile, u/snoyberg.