Nushell core team meeting 2022-10-26

Attendees

• Stefan
• JT
• Reilly
• Darren
• Fernando

Agenda

• dandavison: "I suggest that we start requiring more than one nushell core team member to sign-off on the user-facing changes in a PR. So not necesssarily full code review, but names and signatures of nushell builtins. So maybe it would work by adding to the PR template something saying:
  • "Are you changing a builtin command in a way that affects code written by users? (E.g. adding/removing a command, changing an input/output type, adding a new flag, etc). If so, describe the user-facing changes you're making here. PRs with user-facing changes will require sign-off from at least one other regular contributor before merge."
• Recent PRs
• Experimental parser progress https://github.com/jntrnr/nu_parser2
• Quoting - specifically the last suggestion by dan in this link https://hackmd.io/yBEUx4czTMuHcTICt-xZAA
• Command overloading - keep it or roll it back? Do people have time+energy to polish the feature?
• @nu_shell Twitter access - who wants/needs it?
• Should we remove $nothing?
• Nu Lang Guide
  Image Not Showing Possible Reasons

  • The image file may be corrupted
  • The server hosting the image is unavailable
  • The image path is incorrect
  • The image format is not supported

  Learn More →

Discussed Topics

• Dan's user-facing changes suggestion
  • Darren concerned about adding too much red tape and not being able to land PRs in a reasonable amount of time.
  • Reilly to write up a new PR template draft
• Reducing inconsistencies for users
  • Code duplication with many slightly different decisions
  • Try to provide better documented facilities for implementers in nu-protocol -> single source of truth for common tasks/type coercions
• Quoting
  • Generally like Dan's suggestion
  • Action item: tell Dan, see if he runs with it

PRs

• 6799 fix: fixcd (@Decodetalkers)
  • Windows side lacking, Q: who wants to pick up the tab
  • Not worth discussing yet
• 6809 Add column completion support (@nibon7)
  • Make sure reads have no side effects!
  • command set not yet narrowed down in this PR
  • Probably need to clarify the safe subset of commands
  • Action Item: JT will interact on PR
• 6873 Adds progressbar for 'for' using '-b'. (@FilipAndersson245)
  • Adds dependency on indicatif
  • looks cool, interaction with output unclear
  • not nice as addition on for keyword/command
  • should probably be a command
  • plugin commands as extension
• 6879 make format filesize more flexible (@WindSoilder)
  • Stefan: no opinion on the change yet. code looks complicated with genericism used for one specific case. maybe refactor before landing
  • need an example
  • how will the API change on a larger scale??
  • Afteraction report (stefan): asked for clarification on if this should be a larger refactor. JT also about example/new behavior
• 6885 [WIP] table: Change Record view in expand-mode (@zhiburt)
  • looks good/nice tweak
  • Move along
• 6884 [WIP] table: Show truncated record differently (@zhiburt)
  • before/after
  • Move along
• 6893 make take behave like first (@WindSoilder)
  • BIKESHED ALERT
  • Make a matrix of indexing related commands on what they do and what they operate on
    • Maybe sign up webbedspace for help on that one to record the status quo
    • Make decisions based on that and add the final result to the language guide and release notes
• 6896 Fix ignore-errors for select (@nibon7)
  • Stefan: Don't know if we should silently ignore all errors, -> clarify error handling best practices
  • JT: should this kind of functionality be in a special query language/command? not sure it makes sense in the basic select/get commands
  • points to language guide open questions
• 6904 fix: ✨ "saner" default for mv (@melMass)
  • Stefan: test of denying behavior, release note breaking change, check back with how most shells are doing it.
  • Reilly: LGTM
  • Stefan: needs more tests.
• 6897 Fix each while behavior when printing and maybe in other situations by fusing the iterator (@sholderbach)
  • I need some input how in the world we are able to pull multiple tables from an Iterator + !FusedIterator, would like to add the test or fix the backend later on.

Parser progress

• Experimental parser progress https://github.com/jntrnr/nu_parser2
  Should represent a 0.80 and cover the basic syntax

Interactive mode parser additions should be made optional in a good way

Quoting/string literals

PRs not discussed

Pressing questions to move on those

• 6898 prevent error when lines reads bad bytes (@merelymyself)
  • Stefan: I think should specialize for bytes and not coerce to String, dislike: code duplication, line ending detection

Lightning round

• 6905 Include exit code and command duration in left prompt (@SUPERCILEX)
  • Stefan: purely a question of taste

Larger questions, can we resolve that in this discussion round?

• 6806 Protocol: debug_assert!() Span to reflect a valid slice (@dbuch)
  • Stuck on fixing python virtual env tests. Seems reasonable, should not include user facing changes
• 6798 Skip parsing named flags when not specified (@Yesterday17)
  • larger parser and argument semantics implications
• 6796 Suggestion: input and output syntax shapes (@dandavison)
• 6552 feat(nu-cli): use cwd to filter Hinter suggestions where available (@rrcwang)
  • Stuck on missing tests. Q: are we happy with the configuration options and the defaults.

Seems minor, maybe not worth the larger discussion.

• 6910 Make seq return a ListStream where possible. (@merelymyself)
  • Stefan: sounds like a good idea, needs specialization for ints