Hello, welcome to another weekly log.
I’m continuing to do devops for the Stackage migration. I’m in Phase 3 now, which is the data migration. As mentioned last time, there are Haddocks for nearly every package in every Stackage snapshot in existence living in a bucket, and I need to move that data into a new location. Predictably, I said that the hard part of the Stackage migration is over, and yet I spent most of the day today dealing with some errors I introduced, and I still need to migrate a few processes to use the new bucket. But I’m also continuing to learn a lot about how all the pieces of Stackage work together, and how they work with other parts of the Haskell ecosystem.
Being a good operator means building an intuitive feeling for how properly running system behaves. So I’ve also simply spent time observing the new processes and their usage patterns. Now every time you browse Haddocks on Stackage.org, you will be adding data to the graphs I’m watching.
In other news, GHC meetings have picked up again after the holiday break. So here’s the next dump of GHC issue triage notes.
GHC issue triage notes
-
#24276: The global package db directory of GHC has duplicate files (*.copy) · Issues · Glasgow Haskell Compiler / GHC · GitLab
- bug in bindist Makefile
- probably from change to hadrian
-
#24277: Cannot access cost centre ids through GHC.Stack.CCS interface · Issues · Glasgow Haskell Compiler / GHC · GitLab
- issue with GHC API
- related to ghc internals, which is at the top of ben’s list
-
#24278: GHC Panic when running a test function in ghc-9.6.3 · Issues · Glasgow Haskell Compiler / GHC · GitLab
- may already be fixed, zubin is looking into it
-
#24279: Make Constraint and Type apart, in Core · Issues · Glasgow Haskell Compiler / GHC · GitLab
- Surfaced a type checker bug that unfortunately moves ghc further from the
requested use case - the bug fix will be backported at any rate
- Surfaced a type checker bug that unfortunately moves ghc further from the
-
#24280: Improve performance of `Data.List.sort` by adding a (3,4,5)-way merge · Issues · Glasgow Haskell Compiler / GHC · GitLab
- sounds promising
- there’s still some discussion about performance quibbles
-
#24281: Linear let/where patch is yet to be approved by GHC Steering Committee · Issues · Glasgow Haskell Compiler / GHC · GitLab
- A discussion about the importance of getting GHC changes through the GHC proposal process
-
#24282: SpecConstr regression in NoFib's `spectral/ansi` · Issues · Glasgow Haskell Compiler / GHC · GitLab
- SPJ stuff, over my head
-
#24283: Can GHC and Cabal/Stack co-ordinate so that the GHC project provides an architecture type useful for all? · Issues · Glasgow Haskell Compiler / GHC · GitLab
- “We could, but we shouldn’t”
- Creates a bit too tight of a coupling
-
#24285: (Data.Fixed.mod' n d) not less than d · Issues · Glasgow Haskell Compiler / GHC · GitLab
- “ouch”
- but floats are hard
- it seems like a function with this type simply shouldn’t be expected to be good
-
#24286: -DPROFILING not set when compiling C++ files · Issues · Glasgow Haskell Compiler / GHC · GitLab
- already fixed by Matt
-
#24287: ACQUIRE_LOCK deadlock with ipe · Issues · Glasgow Haskell Compiler / GHC · GitLab
- Takeaway:
barf
is a function that lives up to its name
- Takeaway:
-
#24288: No `MonadFix ((,) a)`? · Issues · Glasgow Haskell Compiler / GHC · GitLab
- basically a CLC proposal draft
-
#24289: Enable documentation building in cross bindists · Issues · Glasgow Haskell Compiler / GHC · GitLab
- related to maybe eventually someday providing cross-compiled bindists, which ghc hq
is definitely not doing any time soon anyway.
- related to maybe eventually someday providing cross-compiled bindists, which ghc hq
-
#24290: cAPI FFI linking issue: Multiple definitions message does not guide towards resolution · Issues · Glasgow Haskell Compiler / GHC · GitLab
- probably not much ghc can do
-
#24291: Missing changelog for (at least) tuples · Issues · Glasgow Haskell Compiler / GHC · GitLab
- actually it’s not certain what the what the problem is. The names changed
-
#24292: Print errors/warnings count as a result of compilation · Issues · Glasgow Haskell Compiler / GHC · GitLab
- is somewhat self-contained
- given some kind of implementation plan, it would be a good self-contained project
- but needs a much more specific specification, which may require some expertise
-
#24293: GHC 9.8 TemplateHaskell lookupValueName for field selector · Issues · Glasgow Haskell Compiler / GHC · GitLab
- type checker stuff going over my head
-
#24294: Lots of incorrect links in haddock of `Data.List.NonEmpty` · Issues · Glasgow Haskell Compiler / GHC · GitLab
- possibly a problem with the hi-haddock patch