-
Notifications
You must be signed in to change notification settings - Fork 31
Miking Meeting Notes 2022
Note-taker: Daniel Lundén
Participants: Daniel Lundén, David Broman, Lars Hummelgren, Anders Ågren Thuné, Gizem Caylak, John Wikman, Linnea Stjerna, Viktor Palmkvist, Oscar Eriksson
Pull requests
- https://github.com/miking-lang/miking/pull/659. We like, merge.
- https://github.com/miking-lang/miking-docker/pull/9. We like, merge.
- https://github.com/miking-lang/miking/pull/660. If the automated tests pass, remove lowering for mi-lite. Then we like, merge.
- https://github.com/miking-lang/miking-emacs/pull/11. Merged already (we like).
Goals for 2023 and beyond (sorted by priority, descending)
- Add proper bootstrapping (https://github.com/miking-lang/miking/issues/624).
- LR parser
- MLang translation
- Dead code elimination
- Set up Miking association.
- Arrange contributor introduction meeting.
- Construct contributor agreement and add contribution guide to website.
- Improve error messages generally.
- Improve documentation (README, website).
- Tutorials
- Official installation page
- Improve handling of externals.
- Add REPL for mi, including work on mi eval.
- Add MExpr with support for imperative constructs (loop, break, return, statements). Investigate join points.
- Add dive operator with partial evaluation.
- Add nested parsing.
Backlog
- Discuss PR merging routines for Miking text editor repositories.
The first Miking meeting for 2023 will take place on January 12, 13:00.
Note-taker: Gizem Caylak
Participants: Viktor Senderov, Gizem Çaylak, Viktor Palmkvist, Lars Hummelgren, John Wikman, Linnea Stjerna, Anders Ågren Thuné, David Broman
Pull requests
- Lars:
-
https://github.com/miking-lang/miking/pull/656
- Test this with miking-dppl
-
https://github.com/miking-lang/miking/pull/656
- Viktor P.:
-
https://github.com/miking-lang/miking/pull/657
- We liked,merge
-
https://github.com/miking-lang/miking/pull/657
- Oscar:
-
https://github.com/miking-lang/miking/pull/658
- We liked,merge
-
https://github.com/miking-lang/miking/pull/658
- Viktor S.:
-
https://github.com/miking-lang/miking-dppl/pull/121
- We liked but it shouldn't be public yet.
-
https://github.com/miking-lang/miking-dppl/pull/121
Discussion
- Miking workshop planning
Note-taker: John Wikman
Participants: Viktor Senderov, Gizem Çaylak, Viktor Palmkvist, Lars Hummelgren, John Wikman, Oscar Eriksson, Linnea Stjerna, Anders Ågren Thuné, David Broman
Pull requests
- Oscar:
- Linnea:
Discussion
- Miking workshop planning
Note-taker: Viktor Palmkvist
Participants: Lars Hummelgren, John Wikman, Oscar Eriksson Pull requests
- Oscar: https://github.com/miking-lang/miking/pull/651
- We like
- Oscar: https://github.com/miking-lang/miking-vscode/pull/24
- We like. We are not sure who is the owner of the repo though, it hasn't been updated for a while.
Discussion:
- On renaming tensors: John would like to avoid adding a reserved name with new syntax on type level.
- Could follow rust:
[Int; 1, 2]
is a two-dimensional array with dimensions 1 and 2. - Discussion around https://github.com/miking-lang/miking/issues/643
- Also connected to refinement types and previous discussion on integer sizes: https://github.com/miking-lang/miking/wiki/Miking-Meeting-Notes-2022#meeting-notes-2022-02-03
- Could follow rust:
Note-taker: Gizem Caylak
Participants: Lars Hummelgren, Gizem Caylak, John Wikman, Daniel Lunden, Oscar Eriksson, Viktor Palmkvist Pull requests
- John:
-
https://github.com/miking-lang/miking-docker/pull/7
- We like and merge.
- Note: @Oscar will add this to the GitHub action.
-
https://github.com/miking-lang/miking-docker/pull/7
- Daniel:
-
https://github.com/miking-lang/miking-benchmarks/pull/45
- We like and merge.
-
https://github.com/miking-lang/miking-benchmarks/pull/45
- Linnea:
-
https://github.com/miking-lang/miking/pull/650
- We like and merge.
-
https://github.com/miking-lang/miking/pull/650
Note-taker: Oscar Eriksson
Participants: Lars Hummelgren, Gizem Caylak, John Wikman, Daniel Lunden, Oscar Eriksson, David Broman, Viktor Palmkvist, Nikolaus Huber
Pull requests
- Viktor:
-
https://github.com/miking-lang/miking/pull/648
- We liked, merge after #644-#647.
-
https://github.com/miking-lang/miking/pull/648
Discussion Lars presented a proof of concept for dynamic loading of externals. It looks promising for the interpreter but a bit more thinking is required for the compilation. There are some issues with dynamic loading and the Owl library. Therefore, projects that depends on this library should consider moving awway from this dependecy.
Note-taker: Gizem Caylak
Participants: Lars Hummelgren, Gizem Caylak, John Wikman, Daniel Lunden, Oscar Eriksson, David Broman, Viktor Palmkvist, William Rågstad, Anders Ågren Thuné
Pull requests
- Daniel:
-
https://github.com/miking-lang/miking/pull/645
- We liked, merge.
-
https://github.com/miking-lang/miking-dppl/pull/118
- We liked, merge. (@Oscar #115 will be closed)
-
https://github.com/miking-lang/miking-benchmarks/pull/44
- We liked, merge.
-
https://github.com/miking-lang/miking/pull/645
- Anders:
-
https://github.com/miking-lang/miking/pull/644
- We liked, merge
-
https://github.com/miking-lang/miking/pull/646
- @Anders will test it and then ping David.
-
https://github.com/miking-lang/miking/pull/647
- This contains previous ones so wait until @Anders ping.
-
https://github.com/miking-lang/miking/pull/644
Note-taker: John Wikman
Participants: David Broman, Lars Hummelgren, Oscar Eriksson, John Wikman, Gizem Çaylak
(No pull requests)
Discussion
- Planning for the Miking release workshop. David has set up a spreadsheet with the tasks to be performed.
Note-taker: Viktor Palmkvist
Participants: Lars Hummelgren, Gizem Caylak, John Wikman, Daniel Lunden, Oscar Eriksson, David Broman, Oscar Eriksson
Pull requests
- Oscar:
-
https://github.com/miking-lang/miking-dppl/pull/115
- We like, but there's a bit more to be done. Further discussions will happen outside of miking meetings, then merge.
-
https://github.com/miking-lang/miking-dppl/pull/115
Discussion
- It would be nice to be able to run inference things and interactively end it, and/or be able to resume an earlier run if you end up needing to run for longer.
Note-taker: Oscar Eriksson
Participants: Lars Hummelgren, Gizem Caylak, John Wikman, Daniel Lunden, Oscar Eriksson, David Broman, William Rågstad, Viktor Palmkvist
Pull requests
- William:
-
https://github.com/miking-lang/miking/pull/642
- We liked, merge.
-
https://github.com/miking-lang/miking/pull/642
Note-taker: Gizem Caylak
Participants: Lars Hummelgren, Gizem Caylak, John Wikman, Daniel Lunden, Oscar Eriksson, Anders Ågren Thuné
Pull requests
- Daniel:
-
https://github.com/miking-lang/miking/pull/640
- Check with Viktor P. for the bindf_ change and then we liked, merge.
-
https://github.com/miking-lang/miking-dppl/pull/112
- First fix and merge #109 with @Lars, after that @Daniel will solve the conflicts (then will ping to merge)
-
https://github.com/miking-lang/miking-benchmarks/pull/43
- We liked, merge
-
https://github.com/miking-lang/miking/pull/640
- Lars:
-
https://github.com/miking-lang/miking-dppl/pull/109
- Presented, and will ping David when it is ready to be merged.
-
https://github.com/miking-lang/miking-dppl/pull/109
Note-taker: Gizem Caylak
Participants: Lars Hummelgren, Viktor Palmkvist, Gizem Caylak, John Wikman, Oscar Eriksson, David Broman
Pull requests
- John:
-
https://github.com/miking-lang/miking/pull/638
- We like, merge.
-
https://github.com/miking-lang/miking-docker/pull/6
- Depends on #638, will be fixed, then John will ping David to be merged.
-
https://github.com/miking-lang/miking/pull/638
- Viktor:
-
https://github.com/miking-lang/miking/pull/639
- Update the description of the flag, then we liked merge.
-
https://github.com/miking-lang/miking/pull/639
Note-taker: Lars Hummelgren
Participants: Lars Hummelgren, Viktor Palmkvist, Daniel Lundén
Pull requests
- Viktor:
-
https://github.com/miking-lang/miking/pull/637
- We like, merge
-
https://github.com/miking-lang/miking/pull/637
- Daniel:
-
https://github.com/miking-lang/miking-dppl/pull/110
- Related issues mentioned in PR can be closed
- We like, merge
-
https://github.com/miking-lang/miking-dppl/pull/111
- We like, merge
-
https://github.com/miking-lang/miking-dppl/pull/110
Note-taker: Oscar Eriksson
Participants: Lars Hummelgren, Gizem Çaylak, Daniel Lundén, David Broman, Lars Hummelgren Online: Anders Ågren Thuné
Pull requests
- Lars:
-
https://github.com/miking-lang/miking/pull/633
- We like, merge.
-
https://github.com/miking-lang/miking/pull/634
- We like, merge.
-
https://github.com/miking-lang/miking/pull/636
- We like, merge.
-
https://github.com/miking-lang/miking/pull/633
- Daniel:
-
https://github.com/miking-lang/miking/pull/635
- We like, merge.
-
https://github.com/miking-lang/miking-dppl/pull/108
- We like, merge.
-
https://github.com/miking-lang/miking-benchmarks/pull/42
- We like, merge. Viktor and/or Linnea might want to have a look.
-
https://github.com/miking-lang/miking/pull/635
Merge order 635, 108, 42, whatever
Note-taker: Lars Hummelgren
Participants: Lars Hummelgren, Gizem Çaylak, Daniel Lundén, David Broman Online: Viktor Palmkvist
Pull requests
- Lars: https://github.com/miking-lang/miking/pull/632
- We like, merge
Discussions
- TyAlls in lambda lifting. Lars will try an approach for this.
- https://github.com/miking-lang/miking-dppl/pull/102 is ready for merge.
Note-taker: Lars Hummelgren
Participants: Lars Hummelgren, John Wikman, Daniel Lundén, Viktor Palmkvist, David Broman Online: Oscar Eriksson
Pull requests
- Viktor: https://github.com/miking-lang/miking/pull/625
- We like, merge
- Lars: https://github.com/miking-lang/miking/pull/629
- Fixes build errors on MacOS, should be merged before other PRs
- We like, merge
Discussions
- We discussed an approach for merging ASTs.
Note-taker: Gizem Caylak
Participants: Lars Hummelgren, John Wikman, Daniel Lundén, Gizem Çaylak
Pull requests General note: there is a problem with the Owl package that causes tests to fail on Mac. Merge PRs after fixing Owl problem.
- Lars: https://github.com/miking-lang/miking/pull/627
- We like, merge
- Daniel: https://github.com/miking-lang/miking-dppl/pull/107
- We like, merge
- Daniel: https://github.com/miking-lang/miking/pull/628
- We close this PR :( (instead optimization will be embedded in miking-dppl/107)
Note-taker: Linnea Stjerna
Online: Linnea Stjerna, David Broman, Oscar Eriksson, Lars Hummelgren, William Rågstad, John Wikman, Viktor Palmkvist, Anders Ågren Thuné, Daniel Lundén, Gizem Çaylak
Pull requests
- Daniel: https://github.com/miking-lang/miking/pull/623
- We like, merge
- Viktor: https://github.com/miking-lang/miking/pull/622 (draft)
- People like the tool but there are drawbacks to the implementation
- Might be a good case study for redesigning language fragments
- For now, we do not merge but leave the PR open
Discussion
- The tuple projection construct in AST builder (
tupleproj_
) generates a non-symbolized name. We see no drawback in symbolizing the name if it introduces no overhead. Viktor P will try out the bootstrapping with and without symbolizing enabled. Long-term policy: the AST builder (and code transformations using them) should maintain the invariant that names are symbolized. - Task force for full bootstrapping (we went through the notes in the open issue). We like the proposal. We agreed that the improved include handling should not allow transitive
include
s. This will mean that we will have to add a lot ofinclude
s in the stdlib files that depend transitively on e.g.bool.mc
right now.
Note-taker: Viktor Palmkvist
Participants: John Wikman, Gizem Çaylak, Daniel Lundén, Viktor Palmkvist, David Broman Online: Linnea Stjerna, Anders Ågren Thuné
Pull requests
- Linnea: https://github.com/miking-lang/miking/pull/621
- We like
- Daniel: https://github.com/miking-lang/miking-dppl/pull/105
- We like
Discussion True bootstrapping task force, not requiring the boot-parser hack
- Goal: implement the current system, with minor improvements, rather than include research efforts
- Current main members: Viktor and John, plan out the tasks some time next week
Note-taker: Oscar Eriksson
Participants: Lars Hummelgren, John Wikman, Gizem Çaylak, Daniel Lundén, Viktor Palmkvist Online: William Rågstad
Pull requests
- Viktor : https://github.com/miking-lang/miking/pull/619
- We like, merge.
- Linnea : https://github.com/miking-lang/miking/pull/616
- Wait until Daniel checks if this PR possibly affects dppl, then we, like merge. Draft PRs
- William : https://github.com/miking-lang/miking/pull/620
- We discussed this PR. William will look at the C external interface and try to make the interface similar to it.
Note-taker: Gizem Caylak
Participants: Lars Hummelgren, John Wikman, Gizem Çaylak, Daniel Lundén, Viktor Palmkvist, David Broman Online: Anders Ågren Thuné, William Rågstad
Pull requests
- Lars : https://github.com/miking-lang/miking/pull/618
- In the documentation, indicate the static check + both flags will complain
- Renaming --debug-accelerate
- Then, we liked merge.
- William : https://github.com/miking-lang/miking/pull/617
- Trampolining by default, two flags : one to trampolining, one to to disable optimizations.
- Move FFI implementation to another PR (to be presented).
- Fix these, no need to review, we liked and ping David then merge.
Note-taker: Gizem Caylak
Participants: Lars Hummelgren, John Wikman, Gizem Çaylak, Daniel Lundén, Viktor Senderov, Anders Ågren Thuné, David Broman
Pull requests
- (DRAFT) Viktor : https://github.com/miking-lang/miking/pull/615
- Not ready to merge yet
- Linnea + Gizem: https://github.com/miking-lang/miking-benchmarks/pull/41
- We liked, merge
Next Miking meeting is planned to be on August 16th.
Note-taker: Gizem Caylak
Participants: Lars Hummelgren, John Wikman, Gizem Çaylak, Daniel Lundén, Viktor Senderov
Online: Anders Ågren Thuné, Linnea Ingmar
Pull requests
- Gizem: https://github.com/miking-lang/miking/pull/615
- We liked, merge
- Linnea: https://github.com/miking-lang/miking/pull/614
- We liked but after merging #613, merge this.
Note-taker: John Wikman
Participants: Lars Hummelgren, John Wikman, Linnea Ingmar, Gizem Çaylak, Daniel Lundén
Online: Anders Ågren Thuné
Pull requests
- Anders: https://github.com/miking-lang/miking/pull/612
- We like.
- Daniel: https://github.com/miking-lang/miking-dppl/pull/100
- We like, but should be merged at the same time as PR #610 since these are related.
- Linnea: https://github.com/miking-lang/miking/pull/613
- We like, though some things Linnea wants to look closer at first. Will ping David when it is ready. Builds on top of PR #610.
Discussion
- Discussed the need to handle the
all
type syntax in the transformations after the type checking. It has been causing some issues for Lars. Various approaches were discussed. Might need to revisit this in the future.
Note-taker: Lars Hummelgren
Participants: Daniel Lundén, John Wikman, Lars Hummelgren, Oscar Eriksson, Linnea Ingmar, Gizem Çaylak
Online: Anders Ågren Thuné
Pull requests
- Daniel
-
https://github.com/miking-lang/miking/pull/609
- We like, merge
-
https://github.com/miking-lang/miking/pull/610
- We like, merge when ready
- Needs to be merged after #608
- Daniel has to make another PR in DPPL before merge
-
https://github.com/miking-lang/miking-dppl/pull/99
- We like, merge
- Should be merged at the same time as #608
-
https://github.com/miking-lang/miking/pull/609
Note-taker: John Wikman
Participants: David Broman, Daniel Lundén, Viktor Senderov, Lars Hummelgren, Gizem Çaylak, John Wikman
Online: Linnea Ingmar, Anders Ågren Thuné
Pull requests
- Linnea: https://github.com/miking-lang/miking/pull/608
- We like, but Daniel will fix some things in miking-dppl that this change will break.
- (The list optimization is an interesting use case for tuning.)
Discussion
- Utest generation with variant types. Lars should be able to fix this issues on his own.
Note-taker: Viktor Palmkvist
Participants: Viktor Senderov, Gizem Çaylak, Oscar Eriksson, David Broman, Linnea Ingmar, John Wikman
Online: Anders Ågren Thuné, William Rågstad, Lars Hummelgren
Pull requests
- William: https://github.com/miking-lang/miking/pull/595
- Included in 607, we want to merge that one instead, so this one should be removed
- William: https://github.com/miking-lang/miking/pull/607
- The tests should be run automatically (compile each file to js, run it with
node
, then run the original file withmi eval
anddiff
the outputs, perhaps via process substitution)- William makes an independent Makefile for the tests, then Oscar helps with integrating it
- Add a
.gitignore
in thejs
folder for the generated.js
files - Then we like
- The tests should be run automatically (compile each file to js, run it with
- William: https://github.com/miking-lang/miking/pull/606
- We like
Note-taker: Oscar Eriksson
Participants: Daniel Lundén, Viktor Senderov, Gizem Çaylak, Oscar Eriksson
Online: Linnea Ingmar, Anders Ågren Thuné, Viktor Palmkvist
Pull requests
- Daniel: https://github.com/miking-lang/miking/pull/605
- Fix type error in PR comment, then we like merge.
- Daniel: https://github.com/miking-lang/miking-dppl/pull/98
- We like, merge.
Note-taker: Gizem Çaylak
Participants: Daniel Lundén, Viktor Senderov, Viktor Palmkvist, Linnea Ingmar
Online: Gizem Çaylak, Anders Ågren Thuné, William Rågstad
Pull requests
- Anders: https://github.com/miking-lang/miking/pull/602
- We like but if you wanna include . tuple projection then fix it in pretty printer
- Anders: https://github.com/miking-lang/miking/pull/604
- We like, merge
- Viktor S: https://github.com/miking-lang/miking-dppl/pull/97
- Unit test fails
- We liked but not ready to merge yet
Note-taker: Viktor Palmkvist
Participants: Daniel Lundén, Viktor Senderov, Gizem Çaylak, David Broman, Lars Hummelgren, Oscar Eriksson
Online: Anders Ågren Thuné, Linnea Ingmar
Pull requests
- Daniel: https://github.com/miking-lang/miking/pull/600
- We like
- Daniel: https://github.com/miking-lang/miking-dppl/pull/96
- The semantics of
Unknown
in the typechecker is currently like anany
, we'd like it to be a type hole, i.e., initially unknown but will be inferred to be some specific type. Anders will look into it. - We like
- The semantics of
- Linnea: https://github.com/miking-lang/trellis-dsl/pull/6
- We like
- Linnea: https://github.com/miking-lang/miking/pull/599
- We like
- Anders: https://github.com/miking-lang/miking/pull/601
- We like
Note-taker: Linnea Ingmar
Participants: Daniel Lundén, Viktor Senderov, Gizem Çaylak, David Broman, Lars Hummelgren, Linnea Ingmar, John Wikman
Online: William Rågstad, Anders Ågren Thuné
Pull requests
- Linnea: https://github.com/miking-lang/miking/pull/598/files
- Change the default prefix to something descriptive, then we like.
- Daniel: https://github.com/miking-lang/miking-dppl/pull/95
- We like, merge.
- Anders: https://github.com/miking-lang/miking/pull/597
- We like, merge.
Note-taker: Gizem Çaylak
Participants: Daniel Lundén, Viktor Senderov, Oscar Eriksson, Lars Hummelgren
Online: Gizem Çaylak, Linnea Ingmar, William Rågstad, Anders Ågren Thuné, Viktor Palmkvist, David Broman
Pull requests
- Daniel: https://github.com/miking-lang/miking-dppl/pull/92
- We liked, merge
- Daniel: https://github.com/miking-lang/miking-dppl/pull/93
- Add test by compiling things
- Rebase since depends on 92, then ping David to merge.
- Daniel: https://github.com/miking-lang/miking-benchmarks/pull/40
- We liked, merge
- William: https://github.com/miking-lang/miking/pull/595
- No merge until Lars' review and failed tests are fixed then ping David.
- Lars: https://github.com/miking-lang/miking/pull/596
- We liked, merge.
Discussion
- Miking meetings are 1-2 pm directly after lunch from now on.
Note-taker: Viktor Palmkvist Participants: John Wikman, Daniel Lundén, Linnea Ingmar, Viktor Senderov, Oscar Eriksson, Lars Hummelgren Online: Gizem Çaylak
Pull requests
- John: https://github.com/miking-lang/miking/pull/592
- We like (but merge after rebase after #593)
- Viktor P & Lars: https://github.com/miking-lang/miking/pull/593
- We like
- Daniel: https://github.com/miking-lang/miking-benchmarks/pull/39
- We like
Discussion
-
https://github.com/miking-lang/miking/pull/590
- Decision on the naming scheme (
errorSingle
,errorMulti
,errorGeneral
), Viktor P fixes, then pings David
- Decision on the naming scheme (
- Thoughts on a different interface for running commands
- No changes now, probably when externals are nicer
- Lars: https://github.com/miking-lang/miking/issues/584
- Some thoughts, nothing conclusive
Note-taker: Viktor S Participants: Anmol D, John W, Daniel L, Gizem C, Linnea I, Viktor S, Oscar E, David B, Lars H Online: Viktor P, William R
Pull requests
- John W, Update the JSON library to not use parser combinators, https://github.com/miking-lang/miking/pull/587
- Linnea I, Add 0-CFA for sequence intrinsics, https://github.com/miking-lang/miking/pull/583
- Linnea I, Add flags for debugging dependency analysis and pretty print optional using in TmUtest, https://github.com/miking-lang/miking/pull/589
- Linnea I, Initial parser for Trellis, https://github.com/miking-lang/trellis-dsl/pull/5
- Viktor P, Typechecking for
tool.mc
, https://github.com/miking-lang/miking/pull/588 - Viktor P, Better errors, https://github.com/miking-lang/miking/pull/590
- Daniel L, Organize CorePPL to MExpr compiler and implement importance sampling, https://github.com/miking-lang/miking-dppl/pull/92
- Daniel L, Mixed additions and updates by dlunde, https://github.com/miking-lang/miking/pull/591
Discussion
- we like #587 miking
- we like #583 miking
- we like #589 mikinig
- we like #5 trellis
- #588 miking introduces
--typecheck
totool.mc
.tpplc
will need to be re-compiled with this flag after merging this PR; we like after tests have passed and ping David afterwards - #590 miking uses Comby tool to rewrite a lot of the code, to make use of the "Better errors";
err
as a top-level name is clashing with what John uses normally, he will provide comments first; we like buterr
needs to be discussed and the previous PR needs to be merged, then ping David - #92 miking-dppl changes to the Makefile for RootPPL, CorePPL, and TreePPL; possibly will conflict, Daniel will resolve.
mapPostAccumL
,mapPreAccumL
, etc. will be moved to the other PR #591 - #591 miking we like, some stuff will have to be moved into it from #92 before merging
Note-taker: Linnea Ingmar
Participants: Lars Hummelgren, John Wikman, Linnea Ingmar, Viktor Palmkvist, Oscar Eriksson, Gizem Çaylak, Viktor Senderov, Anmol Dhawan, David Broman, Daniel Lundén
Pull requests
- Linnea
- https://github.com/miking-lang/miking/pull/583
- Wait with merge, do performance improvements first
- Daniel
- https://github.com/miking-lang/miking-dppl/pull/90
- Ready to be merged, presented previously
- Viktor S
- https://github.com/miking-lang/miking-dppl/pull/91
- Fix type error in the flip model
- Rebase after Daniel's PR is merged
- After that, we like (ping David)
Discussion
- Viktor and Oscar presented the ideas about the new design for externals.
Note-taker: Lars Hummelgren
Participants: William Rågstad, Lars Hummelgren, Anders Ågren Thuné, John Wikman, Linnea Ingmar, Viktor Palmkvist, Oscar Eriksson, Gizem Çaylak, Viktor Senderov, Anmol Dhawan, Daniel Lundén
Pull requests
- Viktor
- Oscar
-
https://github.com/miking-lang/miking/pull/579
- We like, merge
-
https://github.com/miking-lang/miking/pull/579
Discussion
- Daniel will update https://github.com/miking-lang/miking/issues/570 since the mentioned problem is solved by #580 (but not the one mentioned by Linnea).
- Viktor P and Oscar presented proposed approach for externals. More discussions needed.
- Viktor S showed working TreePPL example.
Note-taker: John Wikman
Participants: William Rågstad, Lars Hummelgren, Viktor Palmkvist, Oscar Eriksson, Daniel Lundén, Gizem Çaylak, Viktor Senderov, David Broman, John Wikman.
Pull requests
-
Daniel: https://github.com/miking-lang/miking/pull/576 We like.
-
Daniel: https://github.com/miking-lang/miking-dppl/pull/90 Some rearranging to the src directory to be done before merging. The src directory should follow a flat hierarchy with folders named as "coreppl-to-rootppl" etc. After these changes, we like.
-
Lars: https://github.com/miking-lang/miking/pull/577 We like. Merge PR 576 first, then check if the miking-dppl code works with these changes. Daniel shall ping Lars when 576 is merged, Lars then tests the code and pings David, then David merges.
Discussion
Started discussing https://github.com/miking-lang/miking/issues/562 regarding serialization of data in MCore. Got side-tracked and started talking about the externals system This is important and needed by many. To be discussed further at another meeting.
Note-taker: Oscar Eriksson
Participants: Linnea Ingmar, Gizem Çaylak, Viktor Palmkvist, David Broman, Oscar Eriksson, Lars Hummelgren, Viktor Senderov, John Wikman, Daniel Lundén.
Pull requests
-
John: https://github.com/miking-lang/miking/pull/575 Fix records before merge (John and Oscar). We will use explicit mappings between fields, then merge.
-
Linnea: https://github.com/miking-lang/miking/pull/573 Make
product
useseqLiftA2
, then we like merge.
Discussion
We discussed https://github.com/miking-lang/miking/issues/574. We will leave this issue open.
Note-taker: Daniel Lundén
Participants: Daniel Lundén, Viktor Senderov, Viktor Palmkvist, Oscar Eriksson, John Wikman, Linnea Ingmar, Gizem Caylak, Lars Hummelgren
Pull requests
- https://github.com/miking-lang/miking/pull/572. We like, merge after 569 and 567.
- https://github.com/miking-lang/miking/pull/571. We like, merge after 569.
Discussion
- We discussed problems with unordered and ordered records. Ordered MExpr records are currently used for OCaml externals.
Note-taker: Gizem Caylak
Participants: Lars Hummelgren, Viktor Palmkvist, Oscar Eriksson, Daniel Lunden, Viktor Senderov, John Wikman, Gizem Caylak, Ander Ågren Thuné, William Rågstad
Pull requests
- Linnea: https://github.com/miking-lang/miking/pull/568
- We liked, merge.
Note-taker: Viktor Senderov
Participants: Lars Hummelgren, Viktor Palmkvist, Oscar Eriksson, Daniel Lunden, Viktor Senderov, John Wikman
Discussed PR Re-write OCaml data conversion into a more modular language fragment # 569
We like.
Note-taker: John Wikman
Participants: Lars Hummelgren, Linnea Ingmar, Viktor Palmkvist, John Wikman, Oscar Eriksson
Pull requests
- John
-
https://github.com/miking-lang/miking-docker/pull/5
- Split into baseline and miking images, plus a CUDA image version
- We like
-
https://github.com/miking-lang/miking-docker/pull/5
- Viktor P
-
https://github.com/miking-lang/miking/pull/567
- Make the intrinsics that take two sequences handle the case where their representation differ
- We like (failed tests was due to ocamlformat not being run)
-
https://github.com/miking-lang/miking/pull/567
Discussion
-
https://github.com/miking-lang/miking/issues/566
- Problem with performance if too many temp directories are created due to the suffix index being incremented.
- Agreed upon using a randomized suffix instead of an incremented one.
- Linnea will look into implementing this.
Note-taker: Lars Hummelgren
Participants: William Rågstad, Lars Hummelgren, Linnea Ingmar, Viktor Palmkvist, John Wikman, Oscar Eriksson, Ander Ågren Thuné
Pull requests
- Anders
-
https://github.com/miking-lang/miking/pull/563
- Type checker
- We like
-
https://github.com/miking-lang/miking/pull/563
- John
-
https://github.com/miking-lang/miking/pull/564
- Lambda lifting bugfix
- We like
-
https://github.com/miking-lang/miking/pull/565
- Use
infoErrorExit
in place oferror
where info is available - We like
- Merge after #563
- Use
-
https://github.com/miking-lang/miking/pull/564
Discussion
-
https://github.com/miking-lang/miking/issues/562
- Intrinsic
Byte
data type for representing one byte of data. - Need an efficient way to read/write data from/to disk.
- More discussions needed on this...
- Intrinsic
Note-taker: Viktor Palmkvist
Participants: David Broman, Lars Hummelgren, Anders Ågren Thuné, Daniel Lundén, Viktor Palmkvist, William Rågstad, Viktor Senderov, John Wikman, Linnea Ingmar, Gizem Çaylak
Pull requests
-
https://github.com/miking-lang/miking/pull/557 (William)
- Fix tests, then ping David and we like
-
https://github.com/miking-lang/miking/pull/560 (Lars)
- We like
-
https://github.com/miking-lang/miking/pull/484 (Viktor P)
- Fix tests, remove
str2info
, then ping David and we like
- Fix tests, remove
-
https://github.com/miking-lang/miking-benchmarks/pull/38 (Daniel)
- We like
-
https://github.com/miking-lang/miking/pull/559 (John)
- We like
Note-taker: Gizem Çaylak
Participants: Viktor Palmkvist, John Wikman, Gizem Çaylak, William Rågstad,Viktor Senderov, Daniel Lunden, and Lars Hummelgren
Pull requests
-
https://github.com/miking-lang/miking/pull/557 (William)
- We liked but postpone this to next meeting for David to have a look at it.
-
https://github.com/miking-lang/miking/pull/558 (Viktor P)
- Ping Linnea on tune.mc error emerged with this PR
- Check if it works after after all tests passed then can be merged
-
https://github.com/miking-lang/miking-dppl/pull/89
- We liked, merge.
-
https://github.com/miking-lang/miking-docker/pull/4
- We liked, John will merge it since in miking-docker repo.
Note-taker: Lars Hummelgren
Participants: Daniel Lundén, Lars Hummelgren, William Rågstad, John Wikman, Viktor Palmkvist, Gizem Çaylak, Oscar Eriksson, David Broman, Linnea Ingmar
Pull requests
- John
-
https://github.com/miking-lang/miking/pull/555
- Update microbenchmark and ping David.
-
https://github.com/miking-lang/miking-docker/pull/3
- We like.
-
https://github.com/miking-lang/miking/pull/555
Discussion
- Syntax for expressing types of match argument and return of
sem
.- Add optional type declaration style for
sem
s likesem fn : Int -> () -> ()
, in addition to the existing style. - Empty
sem
s are deprecated:sem fn (a : Int) = -- Intentionally empty
. - Viktor will investigate how much work it would take.
- Accepted formats:
- Add optional type declaration style for
sem leftAllowed_DeclOp : Int -> {parent : ()} -> ()
sem leftAllowed_DeclOp : Int -> {parent : ()} -> ()
sem leftAllowed_DeclOp thing =
sem leftAllowed_DeclOp : Int -> {parent : ()} -> ()
sem leftAllowed_DeclOp (thing : Int) =
sem leftAllowed_DeclOp thing = -- Intentionally empty
sem leftAllowed_DeclOp (thing : Int) =
Note-taker: Oscar Eriksson
Participants: Viktor Palmkvist, Linnea Ingmar, David Broman, John Wikman, Anders Ågren Thuné, Gizem Çaylak, Oscar Eriksson, Viktor Senderov, Daniel Lunde, and Lars Hummelgren
Pull requests
-
https://github.com/miking-lang/miking/pull/552 (Lars)
- We like, merge.
-
https://github.com/miking-lang/miking/pull/555 (John)
- Do not include in the standard library but add as micro benchmark.
-
https://github.com/miking-lang/miking/pull/556 (Viktor)
- We like, merge.
Discussion Explicit or implicit universal quantification (based on this issue https://github.com/miking-lang/miking/issues/554). Viktor presented the problem and we decided on fully explicit quantification.
Note-taker: John Wikman
Participants: Viktor Palmkvist, Linnea Ingmar, David Broman, John Wikman, Anders Ågren Thuné, Gizem Çaylak, Oscar Eriksson
Pull requests
- Anders: https://github.com/miking-lang/miking/pull/553
- We like (after the errors have been fixed)
- Some additional things to be done, but they can be left to a future PR
Discussion
- Discussed whether or not to force explicit for-all quantification on annotations of unbound types. Arguments were brought up for and against. Viktor presented an example where it might be good to have explicit for-all quantifiers in the type annotation. He will create an issue in GitHub about this where the discussion can be had further.
Note-taker: Viktor Palmkvist
Participants: Daniel Lundén, John Wikman, Gizem Caylak, Linnea Ingmar, Viktor Palmkvist, Lars Hummelgren, Anders Ågren Thuné, William Rågstad, Oscar Eriksson
Pull requests
- Lars: https://github.com/miking-lang/miking/pull/550
- We like
- John: https://github.com/miking-lang/miking/pull/551
- Confirm that we can pprint these as well, then we like
- There's a test failure caused by one of the underlying actions failing, so the automatic tests need to be rerun, but only David can do so
- John: https://github.com/miking-lang/miking-docker/pull/1
- We like
Note-taker: Daniel Lundén
Participants: Daniel Lundén, Lars Hummelgren, Gizem Caylak, Linnea Ingmar, Oscar Eriksson, John Wikman, William Rågstad, Viktor Palmkvist
Pull requests
- https://github.com/miking-lang/miking/pull/549 We like, merge.
Discussion
- Viktor presented his work on the parsing framework.
Note-taker: Gizem Caylak
Participants: Linnea Ingmar, Daniel Lundén, Gizem Caylak, Anders Agren Thune, David Broman, Viktor Palmkvist
Pull requests
- Anders: https://github.com/miking-lang/miking/pull/510
- We liked, merge.
- Lars: https://github.com/miking-lang/miking/pull/543
- We liked, merge.
Discussion
- Lars will make an urgent PR on dppl crush and Daniel will check it.
- Linnea will change the documentation and add (some) PR to fix opam pin.
Note-taker: Linnea Ingmar
Participants: Linnea Ingmar, Daniel Lundén, Gizem Caylak, Oscar Eriksson, David Broman, Viktor Palmkvist
Pull requests
- Daniel: https://github.com/miking-lang/miking-dppl/pull/84
- We like, merge.
- Daniel: https://github.com/miking-lang/miking-benchmarks/pull/37
- We like, merge.
- Viktor: https://github.com/miking-lang/miking/pull/544
- We like, merge.
- Lars: https://github.com/miking-lang/miking/pull/543
- We suggested using the info format (same as for errors) in the output (use
info2str
for output). - After that, we like.
- We suggested using the info format (same as for errors) in the output (use
Discussion
- We discussed new agreements about the office (when to be there and who sits in which room).
- Next time we'll discuss Viktor's parser generator DSL.
Note-taker: Lars Hummelgren
Participants: Lars Hummelgren, Linnea Ingmar, Anders Ågren Thuné, Daniel Lundén, Gizem Caylak, Oscar Eriksson
Pull requests
- Daniel
-
https://github.com/miking-lang/miking-dppl/pull/87
- Updated to work with https://github.com/miking-lang/miking/pull/540
- We like, merge (also merge ^)
- https://github.com/miking-lang/miking-dppl/pull/84
-
https://github.com/miking-lang/miking-benchmarks/pull/37
- Wait with these
-
https://github.com/miking-lang/miking-dppl/pull/87
Discussion
- Utests: https://github.com/miking-lang/miking/issues/541
- Lars will add file name to output of failed utests
Note-taker: Gizem Caylak
Participants: Lars Hummelgren, Gizem Caylak, Linnea Ingmar, Oscar Eriksson, David Broman
Pull requests
- Lars: https://github.com/miking-lang/miking/pull/540
- We liked but wait for now: Lars removes command line part and talks to Daniel for dppl part.
Note-taker: Daniel Lundén
Participants: Daniel Lundén, Anders Ågren Thuné, Lars Hummelgren, Gizem Caylak, Linnea Ingmar, Oscar Eriksson
Issues
- We updated and removed a few
miking
issues.- Closed https://github.com/miking-lang/miking/issues/45.
- Updated https://github.com/miking-lang/miking/issues/100 with new suggestion.
- Closed https://github.com/miking-lang/miking/issues/102.
- Closed https://github.com/miking-lang/miking/issues/181 (Commenting format now added to conventions in wiki). Note: Viktor or David must close this.
- Closed https://github.com/miking-lang/miking/issues/357.
- Closed https://github.com/miking-lang/miking/issues/410.
Note-taker: Viktor Palmkvist
Participants: Lars Hummelgren, Gizem Caylak, Daniel Lunden, Linnea Ingmar, David Broman, William Rågstad.
Pull requests
- Lars: https://github.com/miking-lang/miking/pull/538
- We like
Discussion
-
Lars: how to control the size of integers/floats in MExpr? Lots of back and forth discussion on what to do.
Concrete conclusions:
- Accelerate will choose size through a flag, using 64 bit as the default size.
- Default ints in MExpr are at least 53 bits, with undefined behavior if a value gets too large to fit.
Other thoughts and relevant things:
- A new kind of expression that specifies the minimum size required for the contained expression, using some form of dynamic scoping (i.e., it's not just the precise expression, but also functions called from there etc.). These would nest by choosing the smallest size sufficient for all such constraints that are "in scope". However, this would most likely not affect things that are not under any constraint, these would instead use the 53 bit default size.
- We could have a subset of MExpr that is extremely explicit in sizes, including conversion functions etc. Programs would be compiled from something more high level into MExpr that uses something more convenient (the previous point or something else), then transformed into something that only uses these extremely specific types, presumably by a pass that knows what backend we're compiling towards.
- Attach possible range information to an int type, propagate appropriately through arithmetic and the like, then provide explicit
wrap
andcheck
operations for getting the range down to a smaller range. The compiler would choose a backing type based on what's required, and choose appropriate instructions for the combination ofwrap
/check
and arithmetic being done. Inspiration: https://blog.polybdenum.com/2021/10/03/implicit-overflow-considered-harmful-and-how-to-fix-it.html - One relevant point is that it may be useful to switch precision in a quite granular manner: https://ieeexplore.ieee.org/abstract/document/8443735
Note-taker: Oscar Eriksson
Participants: Linnea Ingmar, Lars Hummelgren, Gizem Caylak, Viktor Palmkvist, David Broman, Oscar Eriksson.
Pull requests
- Lars
-
https://github.com/miking-lang/miking/pull/536
- We like, merge.
-
https://github.com/miking-lang/miking/pull/537
- We like, merge.
-
https://github.com/miking-lang/miking/pull/536
Discussion We again discussed the semantics of sequence intrinsic (summarized here https://github.com/miking-lang/miking/issues/531). The conclusion is that we like it but it needs more testing.
Note-taker: Gizem Caylak
Participants: Linnea Ingmar, Lars Hummelgren, Gizem Caylak, Viktor Palmkvist
Pull requests
- Linnea
-
https://github.com/miking-lang/miking/pull/535
- We liked, merge.
-
https://github.com/miking-lang/miking/pull/535
Discussion
- We discussed https://github.com/miking-lang/miking/issues/531 and came to a suggestion but suggestion should be made an issue.
Note-taker: Daniel Lundén
Participants: Daniel Lundén, Viktor Palmkvist, Anders Ågren Thuné, Lars Hummelgren, Gizem Caylak, Linnea Ingmar, Oscar Eriksson, David Broman
Pull requests
-
https://github.com/miking-lang/miking/pull/533. We should make the
end
when defininglang
s compulsary. Lars adds this to the parser and fixes any remaining parse conflicts. Lars also adds type checking of certain files as part oftest-all
to test the new functionality introduced by the PR. Viktor will also check the things Lars adds. - https://github.com/miking-lang/miking/pull/534. We like. Merge after Viktor adds some tests.
- https://github.com/miking-lang/miking/pull/526. We like. Merge after Linnea updates the PR description slightly and adds a microbenchmark.
Discussions
- We discussed compiler performance issues discovered by Linnea and Lars.
- We discussed adding unique symbols to the AST as a substitute for ANF (which is currently used for this purpose).
Note-taker: Gizem Caylak
Participants: Linnea Ingmar, Lars Hummelgren, Oscar Eriksson, Gizem Caylak, Daniel Lundén
Pull requests
- Gizem
-
https://github.com/miking-lang/miking/pull/528
- We like, merge.
-
https://github.com/miking-lang/miking/pull/528
- Lars
-
https://github.com/miking-lang/miking/pull/533
- We liked but let Viktor have a look at it.
-
https://github.com/miking-lang/miking/pull/533
Discussion
- It would be good if everyone read the issue https://github.com/miking-lang/miking/issues/531 and we can discuss it later
Note-taker: Lars Hummelgren
Participants: Linnea Ingmar, Lars Hummelgren, Viktor Palmkvist, Oscar Eriksson, Anders Ågren Thuné, David Broman, Daniel Lundén
Pull requests
- Linnea
- Daniel
-
https://github.com/miking-lang/miking-dppl/pull/82
- Fix return value in
main.mc
when given empty arguments. - Then we like.
- Fix return value in
-
https://github.com/miking-lang/miking-benchmarks/pull/36
- We like.
-
https://github.com/miking-lang/miking-dppl/pull/82
Note-taker: Viktor Palmkvist
Participants: Viktor Palmkvist, Anders Ågren Thuné, Lars Hummelgren, Gizem Caylak, Linnea Ingmar, David Broman, Oscar Eriksson
Pull requests
- Lars: https://github.com/miking-lang/miking/pull/520
- This discovers some undefined behavior in the compiler. Lars will write up the possible edge cases and bring up as a discussion for a later meeting.
- We like
- Lars: https://github.com/miking-lang/miking/pull/525
- We like
- Lars: https://github.com/miking-lang/miking/pull/527
- We like
- Oscar: https://github.com/miking-lang/miking/pull/524
- We like
- Linnea: https://github.com/miking-lang/miking/pull/529
- We like
Discussion
- Viktor: create an issue to rename the
smap
functions to not use snake case.
Pull requests postponed