arybczak / list-tries Goto Github PK

https://hackage.haskell.org/package/list-tries-0.6.0/reports/1

i think that 0.6 is still < 7.10, i'll use my trustee powers to limit it for now. https://hackage.haskell.org/package/list-tries-0.6.0/revisions/ (I / you can always revert those bounds, but seems like the right thing to do )

you should also add a .travis.yml and try out travis CI for easily testing the range of GHC's you wish to support before pushign to hackage! :)

https://github.com/hvr/multi-ghc-travis is a good template to use
may OSS haskell projects use that or some customization thereof, eg my hblas package https://github.com/wellposed/hblas/blob/master/.travis.yml
or the lens project https://github.com/ekmett/lens/blob/master/.travis.yml both use it in some form

Can you please relax the base upper bound on Hackage to < 4.13 to allow building with GHC 8.6?

Similar to how Foldable (m k) is a superclass, it seems like Functor (m k) would also be a useful superclass instead of requiring a custom map function.

It is not clear which one to use in which situations.

Here is maybe a relevant quote from stackoverflow:

Oh, and since the package didn't state this explicitly anywhere I could see: The "Patricia trie" version is a trie that compresses sequences of single-branch nodes into a single node that stores the common key prefix. So for keys "aabb" and "aabc" you'd get a node with "aab" and then branches "b" and "c". The standard trie always branches one element at a time.

-- Generic Trie Haskell implementation

Unfortunately I'm fairly out of touch with the Haskell community these days and don't really have the time or interest to maintain either https://github.com/Deewiant/list-tries or https://github.com/Deewiant/glob properly. For example, I used to be proactive about following the Travis CI notifications about GHC HEAD failing and fixing issues before the next GHC release comes out, but these days I'm letting things languish and only react when someone opens an issue or PR about it. Let alone actually doing more development than the minimal compatibility fixes.

So I'm facing the facts and willing to relinquish ownership of both repos to someone more active. If you're interested, let me know.

Equivalent issue for Glob: Deewiant/glob#36

Do (or could) any of the data structures in this package support longest-prefix lookup? That is, if my key is [a,b,c,d,e] and the map has an entry with key [a,b,c] (but not [a,b,c,d] or [a,b,c,d,e]) I'd like a lookup function that returns the value for [a,b,c]. I did not see that listed in the docs, but it seems like a very natural query for maps with lists as keys...