{"thread":{"id":"66794","subject":"[Outreachy] column: propose using early_scan_options() for --command","startedAt":"2026-11-06T17:26:48Z","lastAt":"2026-11-06T17:26:48Z","messageCount":1,"participants":["Fien Dora"],"isPatch":false,"patchVersion":null,"patchTotal":null},"messages":[{"id":"557717","messageId":"CACWAS-X1EGU+esDOeym+HnLAUqyBO3nBsboPU0RTV-RF1HKDfA@mail.gmail.com","threadId":"66794","inReplyTo":null,"subject":"[Outreachy] column: propose using early_scan_options() for --command","fromName":"Fien Dora","fromEmail":"fiendora5@gmail.com","sentAt":"2026-11-06T17:26:48Z","receivedAt":"2026-11-06T17:26:48Z","isPatch":false,"sender":{"key":"fiendora5@gmail.com","avatar":null},"body":"Hello Christian, Siddharth, and the Git community,\n\nFollowing my earlier contribution, t1402: test --no-allow-onelevel\noption precedence, I have been exploring possible follow-up work\nrelated to Git's argument-scanning and parsing improvements.\n\nI have been studying the new early_scan_options() API, particularly\nits implementation in parse-options.c and its use in fast-import.\n\nI noticed that builtin/column.c still performs a manual early check\nfor --command= before invoking parse_options():\n\nif (argc > 1 && starts_with(argv[1], \"--command=\"))\n\nThis is necessary because git column uses the option's value to load\nconfiguration before normal option parsing. However, the early check\nrecognizes only the exact --command=<name> form, whereas\nparse_options() also recognizes separate-value and unambiguous\nabbreviated forms.\n\nI tested the current implementation and observed that\n•⁠  ⁠--command=branch is accepted.\n•⁠  ⁠--command branch and --comm=branch are rejected.\n•⁠  ⁠--command=branch is rejected when preceded by another option.\n•⁠  ⁠Repeated identical --command values are accepted, while\nconflicting values are rejected.\nI would like to propose a focused follow-up patch that replaces the\nmanual early check with early_scan_options(), mark --command with\nPARSE_OPT_EARLY, and add regression tests for the relevant\nargument-handling behavior.\n\nI would appreciate guidance on whether the existing first-position and\nattached-value restrictions should be preserved or whether the\nconversion should accept additional forms recognized by\nparse_options().\n\nI also noticed that Cynthia Uche mentioned git column as a possible\nconversion target in her internship proposal. I would like to\nacknowledge her proposal to prevent any potential overlap before\nstarting any implementation.\n\nWould this be a suitable independent follow-up contribution to\nundertake now, or would you recommend a different target?\n\nThank you for your time and guidance.\n\n"}]}