Re: [PATCH v3 00/12] fast-import: standardize usage string and SYNOPSIS
Elijah Newren <[email protected]>
| Newsgroups | org.kernel.vger.git |
|---|---|
| Message-ID | <CABPp-BH4MmEL5vZv7ZoUufT8quk4vCq==Ef_XAfxjj9U-WBsEw@mail.gmail.com> |
On Tue, Aug 11, 2026 at 1:33 AM Christian Couder <[email protected]> wrote: > > Changes since v2 > ================ > > Thanks to Elijah for reviewing v2 and Junio for reviewing v1. > > The series has been rebased on top of recent master at 010afd3166 (The > 12th batch, 2026-08-07). > > - Patch 2/12 now describes `PARSE_OPT_LASTARG_DEFAULT` the way Elijah > suggested, which matches what "parse-options.h" says, and mentions > that it should not be combined with `PARSE_OPT_OPTARG`. > > - Patch 7/12 now wraps the function signatures that grew too long > when they gained a 'struct fast_import_state *state' argument, with > one parameter per line. Patches 8/12 and 12/12 have been adjusted > accordingly, as they touch two of these signatures again. > > - Patch 11/12 now also lists two other behavior changes that Elijah > spotted: value-taking options accept the space-separated > `--opt value` form, and a bare or trailing `--` is now accepted. > > - Patch 11/12 also documents, both in a NEEDSWORK code comment and > in "Documentation/git-fast-import.adoc", the third behavior change > Elijah spotted about `--allow-unsafe-features`: as the early scan > for that option only matches its exact spelling and stops at the > first non-option argument, it disagrees with parse_options() for > command lines like `--allow-unsafe` or > `--depth 5 --allow-unsafe-features`. This errs on the safe side, > as unsafe `feature` commands from the stream are refused in that > case. > > I plan to work on a follow-up series that will improve on this by > teaching the early scan about the options that take a value. It > looks like the parse-options API could provide some helpers for > this, which might benefit other commands like `git` itself (see > handle_options() in "git.c") that need to look at some options > before the actual option parsing. Thanks, this version addresses all my feedback on the previous round, and I spotted no new issues introduced. Looks good to me.