Skip to content
GitLab
Projects Groups Snippets
  • /
  • Help
    • Help
    • Support
    • Community forum
    • Submit feedback
    • Contribute to GitLab
  • Sign in / Register
  • S stdpp
  • Project information
    • Project information
    • Activity
    • Labels
    • Members
  • Repository
    • Repository
    • Files
    • Commits
    • Branches
    • Tags
    • Contributors
    • Graph
    • Compare
  • Issues 81
    • Issues 81
    • List
    • Boards
    • Service Desk
    • Milestones
  • Merge requests 12
    • Merge requests 12
  • CI/CD
    • CI/CD
    • Pipelines
    • Jobs
    • Schedules
  • Deployments
    • Deployments
    • Releases
  • Analytics
    • Analytics
    • Value stream
    • CI/CD
    • Repository
  • Wiki
    • Wiki
  • Activity
  • Graph
  • Create a new issue
  • Jobs
  • Commits
  • Issue Boards
Collapse sidebar
  • Iris
  • stdpp
  • Issues
  • #64
Closed
Open
Issue created May 20, 2020 by Paolo G. Giarrusso@BlaisorbladeContributor

Further issues with in-Coq computation for pmap and gmap

These issues are discussed on Mattermost starting at https://mattermost.mpi-sws.org/iris/pl/o9tme6imrfdtf8chhwi6fr1kor. Basically, reducing gmaps will also reduce their well-formedness checks; I expect continuing the work in !106 (merged) and sealing well-formedness predicates would avoid the issue.

Of most interest are these comments:

Let me point out that computing gmap_wf_aux takes time linear in the sum of key sizes, and that plausibly explains the performance numbers that I saw back then, and the problem Greg mentions with large keys. Meanwhile, Pmap_wf has a smaller variant of the same problem (its cost does not depend on key sizes).

and Robbert's idea for Pmap:

Also, at some point I looked into a more intrinsic representation of finite maps that ensures the well-formedness condition without a Sigma-type. I have never finished that, but if someone wants to help working on this, feel free to get in touch.

The code is in https://mattermost.mpi-sws.org/iris/pl/6mkua571y7g35cqndt455y74rc; I expect some sealing would still be needed.

Assignee
Assign to
Time tracking