(952a9f8) Complete the verb, the statuses, the tags and the campaigns in the prompt, and bump to 0.33.4

With the ticket suggested, the rest of the line was still typed blind. The
prompt now completes everything fish does: the verb and the verbs of a
group, the five statuses, the tags the board carries, the campaigns a
codename resolves against, and relate --clear's three kinds.

`ticket_argument` becomes `argument_at`, which answers what a position
names rather than whether it is a ticket. The tables say Ticket, Campaign
or Tag; clap says the rest, so the verbs, the statuses, `set --clear`'s
fields and the `completions` shells are never written down twice. Only
`relate --clear` needed a table of its own: the flag takes a kind and then
a target, so possible values on it would refuse every target.

The row draws the longest opening every candidate agrees on, which is
exactly what the next tab takes. One rule for one candidate and for many: a
ticket position has only the selected card, so the rest of its id is drawn
as before; the five statuses agree on nothing, so nothing is drawn and tab
steps through them one press at a time.

The walk holds the line as it stood before the first press, not the list
and not the candidate the last press chose, so every press re-derives the
candidates and stepping cannot narrow them to what it just typed. Every
edit ends it, beside `refused` and for the same reason.

193f9be is not inherited: `tags set` is offered only the tags the
vocabulary declares, and `tags add` and both `--tags` that take the whole
list are offered nothing.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
This commit is contained in:
Ansho Rei
2026-09-15 15:18:59 +02:00
co-authored by Claude Opus 5
parent c714c73a75
commit a885f81da5
6 changed files with 697 additions and 113 deletions
+3 -1
View File
@@ -13,8 +13,10 @@ The generated half, the fish half, and what keeps the two honest.
- **The completion lists are kept honest by tests** that walk the clap definition and fail on an argument left out of them. `TICKET_COMMANDS` and `TICKET_SECOND_COMMANDS` by position, `TICKET_FLAGS`, `CAMPAIGN_COMMANDS`, `CAMPAIGN_NESTED_COMMANDS`, `CAMPAIGN_FLAGS` and `TAG_FLAGS` likewise, the last against the test-local `TAG_LIST_FLAGS` and `TAG_NEW_COMMANDS` for the tag arguments that deliberately get no rule. The walk is one function, `every_argument`, since the tests differ only in the id they look for. Each also asserts *how many* it found, so a walker that stopped finding arguments fails rather than passing every assertion by never running them
- **A campaign is named in three shapes and a tag in one**, which is why the lists are not one list: a campaign is `assign`'s first positional, the first argument of `campaign show` and `campaign set`, and `ls --campaign`'s value, and each needs a different fish rule. A tag is only ever an option value, and its values come from a different command
- **`TICKET_FLAGS` holds pairs, the command and the flag, and every ticket rule in the fish half is generated from it.** Three commands carry a `--clear` and each takes something different: `chown`'s names the ticket, since with no user to give there is no positional for it; `relate`'s takes the kind and then the target; `set`'s takes a field name clap already offers. So the command is half of the fact, and a list of bare flag names could only be scoped by hand, which is what the `chown --clear` rule used to be. `every_ticket_flag_is_a_real_option` walks each pair against the command it names, which a flat list could not do
- **The lists answer the board as well as the shell.** `ticket_argument` reads the three ticket lists to say whether the argument at a given index of a typed line is a ticket, which is what the command prompt's suggestion is built on, see [tui-prompts.md](tui-prompts.md). Position is counted there exactly as it is counted here, so the two cannot come to disagree about `mv --status done <ticket>`; a second table on the board would be a copy the tests above never read
- **The lists answer the board as well as the shell.** `argument_at` reads every one of them to say what the argument at a given index of a typed line names — a ticket, a campaign, a tag, or a fixed list it takes off clap — which is what the command prompt completes against, see [tui-prompts.md](tui-prompts.md). Position is counted there exactly as it is counted here, so the two cannot come to disagree about `mv --status done <ticket>`; a second table on the board would be a copy the tests above never read. The values are the one half that is not shared: fish runs `worklane ls --ids` and the board reads the rows the tab already holds
- **`TAG_FLAGS` holds pairs where the others hold names**, because the two options that take one tag are spelled differently: `ls --tag` filters on one and `append --tags` adds them, so a single rule naming both would tell fish that each command accepts the other's flag. One rule per pair, generated in `fish_dynamic`. `add --tags` and `set --tags` are in the test's `TAG_LIST_FLAGS` instead and get no rule: they take the *whole* list, so the token at the cursor is `bug,tui` and fish replaces a token entire, which would throw away what is already typed. The test walks both `tag` and `tags`, so the next one has to say which of the two it is
- **A `ValueEnum` is never listed twice, and `argument_at` is why the fish half's two hand-written value lists are the only ones.** clap emits the possible values of an *option*, so `mv --status` and `set --clear` need nothing here and the board reads them off the same definition with `Arg::get_possible_values`. The `ls` status and the `completions` shell are positionals, which the fish generator emits nothing for, so those two are written into the block from the enums — and the board reaches them through `get_positionals`, which needs no `Command::build` the way `get_index` would
- **`RELATION_KIND_FLAGS` is the one position clap cannot answer for.** `relate --clear` takes a kind and then a target, two values of different kinds under one flag, so possible values on it would refuse every target. Both renderers therefore say so themselves: the fish rule lists `RelationKind::ALL` and the table names the pair
- **`ls --ids` is the machine-readable listing**, and one of the three outputs in the tree meant to be parsed: the short id, a tab, the title, unstyled. Two fields because fish reads what follows a tab as the description and a script reads `cut -f1`. It includes done, which the decorated listing leaves out, because a completion has to be able to name a finished ticket, and it skips a missing status directory rather than refusing, because an error printed over a shell prompt is worse than no candidates. The filters still apply, and an unknown `--campaign` is still refused
- **`campaign ls --ids` is the same flag and the same two fields for the other object**, and it deliberately skips the member counting the decorated listing does: `members()` reads every ticket on the board, and the caller here is a shell filling in candidates on a keypress. It also prints nothing on an empty board, where the decorated listing prints a sentence, because a sentence offered as a candidate is worse than no candidates
- **`worklane tags` is the third, and it is a command rather than a mode on `ls`.** Two reasons: `ls --tag <tag>` answers what carries a tag and cannot answer what tags exist, which is the question a grooming sweep starts from, and a `--tags` beside a `--tag` is one letter from a filter that means something else. Alphabetical and case sensitive, matching the filter, so `bug` and `Bug` list separately and land next to each other, which is where a near-duplicate is spotted. Every status, done included: a tag that has fallen out of use is still part of what the vocabulary has been. `assets/PLAN.md` used to tell the agent to count tags with a `show` per ticket, and now points here
+13 -4
View File
@@ -46,16 +46,25 @@ The three input modes on the board and the key that copies an id off it.
- **A failed write is silent, and an unreadable file is an empty recall.** The line is in memory either way, so the walk works for the run on a board whose folder cannot be written, and a board with no history yet is the ordinary case rather than something to say. A notice about a history file would be about the one thing the reader did not ask for
- **`.wlhistory` is not in `digest`.** Appending to it must not read as the tree moving, or every command typed would provoke a reload of every board on top of the one `run` already does
## The Prompt's Suggestion
## The Prompt's Completion
- **The reader has the ticket selected, which is the whole reason to type a command on the board rather than in a shell.** So where the line is naming a ticket, the selected card's short id is drawn after the cursor and `tab` takes it into the line. Retyping seven characters off the card in front of you is the friction the prompt exists to remove
- **Where a ticket goes is `completions.rs`'s answer, not a table written here.** `ticket_argument` reads `TICKET_COMMANDS`, `TICKET_SECOND_COMMANDS` and `TICKET_FLAGS`, which `no_ticket_taking_subcommand_is_missing` and `every_ticket_flag_is_a_real_option` keep honest against clap's own definition. A second copy of the positions on the board would go wrong the next time a command grew a ticket argument, and it would go wrong *silently*, since the test guarding the first copy would not read the second
- **Where a ticket goes is `completions.rs`'s answer, not a table written here.** `argument_at` reads `TICKET_COMMANDS`, `TICKET_SECOND_COMMANDS` and `TICKET_FLAGS`, which `no_ticket_taking_subcommand_is_missing` and `every_ticket_flag_is_a_real_option` keep honest against clap's own definition, and the campaign and tag lists beside them. A second copy of the positions on the board would go wrong the next time a command grew a ticket argument, and it would go wrong *silently*, since the test guarding the first copy would not read the second
- **The ticket is not always the first argument, which is what a naive version gets wrong.** `chown alice <ticket>` and `assign v1 <ticket>` put it second on purpose, `relate x --blocked-by <ticket>` puts it in a flag, and `chown --clear <ticket>` does too, since with no user to give there is no positional for it. Nothing is offered where the argument is something else: `relate --clear` wants a kind, `set --clear` wants a field name, and `chown`'s first argument is a user
- **Position is counted, exactly as the fish half counts it**, so `mv --status done <ticket>` is not recognised: the ticket is `mv`'s first positional and the third argument on the line. That is the fish rules' limit as well, and one table answering both is worth more than two tables disagreeing about the corner
- **`TICKET_FLAGS` holds pairs, command and flag, where it used to hold bare flag names.** Three commands carry a `--clear` and only `chown`'s names a ticket, so the command is half of the fact. It also took the last hand-written ticket rule out of the fish block: every one of them is generated from the table now, and `every_ticket_flag_is_a_real_option` walks each pair against the command it names
- **It is derived on every call and held nowhere.** `Board::suggestion` is asked once by the draw and once by `tab`, so what is on the row and what the key takes cannot disagree, and there is no field to go stale when the line is edited or the selection moves. That is also what "any other key drops it" means: nothing is dropped, because nothing was kept
- **It is derived on every call and held nowhere.** `Board::completion` is asked once by the draw and once by `tab`, so what is on the row and what the key takes cannot disagree, and there is no list to go stale when the line is edited or the board reloads under it. That is also what "any other key drops it" means: nothing is dropped, because nothing was kept
- **Only with the cursor at the end of the line.** The suggestion is drawn after the cursor and `tab` inserts it there, so with a command continuing past the cursor both would land in the middle of something already typed. A shell's own suggestion keeps the same rule
- **Only while the selected id still starts with what is typed**, and what is drawn is the rest of it rather than the whole. A suggestion that no longer matches the argument under it describes nothing, and once the whole id is typed there is nothing left to add, so a second `tab` cannot double it
- **Only while a candidate still starts with what is typed**, and what is drawn is the rest of it rather than the whole. An offer that no longer matches the argument under it describes nothing, and once the whole of it is typed there is nothing left to add
- **Nothing with no card under the selection.** On the frame the selection is on a campaign or a tag, and neither is a ticket: `y` copies what the frame names there, and this deliberately does not
- **Everything else worklane can name is completed the same way**, which is what makes it one mechanism rather than a ticket suggestion with four more bolted on. `Argument` says what a position takes — a ticket, a campaign, a tag, or a fixed list — and `Board::candidates` says what the board currently holds of it. The verbs and the enum values are clap's own, so a command, a status, a shell or a clearable field added anywhere is typeable here the day it is added, and the campaigns and the tags are the tab's own rows, rebuilt at every reload beside the columns and drawn by the `c` and `t` pickers. **Nothing here reads the tree**: a keypress that went to disk would be a second source for what is already in memory
- **A campaign is offered by codename, and by its short id where it has none.** The codename is what the board draws on every card in the campaign and the shortest thing that addresses it; a campaign without one is labelled by its short id in both renderers, and leaving it out would hide it from the only place a reader can reach it by typing
- **`tags set` is offered only the declared tags.** It changes what a declared tag means and refuses one the vocabulary file does not declare, so the carried-but-undeclared ones are exactly the values it rejects. That is 193f9be in the fish half, which reads one listing for both, and the prompt does not inherit it. `tags add` gets nothing for the mirror reason, and neither `--tags` that takes the whole list does, since one tag offered into `bug,tui` is not what the field asked for
- **The clap definition is rebuilt on every call**, at 32µs in release against a board that redraws four times a second. A definition cached beside the board would be a second copy of what `Cli::command()` already answers; the fish half spends 2.5ms per keypress running worklane itself for the same values
- **What is drawn is the longest opening every candidate agrees on, which is exactly what the next `tab` takes.** One rule for one candidate and for many: a ticket position has only the selected card, so the rest of its id is drawn; the five statuses agree on nothing, so nothing is drawn and `tab` steps through them instead. The row never shows something a press would not do
- **`tab` again steps through the candidates in place, and wraps.** A shell does the same and it needs no list drawn over the columns, which is the trade — the board behind the prompt is what the reader is reading. Wrapping loses nothing typed, unlike the recall's walk, because every candidate still starts with the stem the walk was built from, which is why there is no `pending` here
- **Each press rebuilds the row from the line as it stood before the first**, so the candidates are read fresh every press and the only thing `Completing` holds is that line and how many presses have landed. Holding the list would be a copy going stale between two presses; holding the candidate the last press chose would narrow the list to it, and the walk would step once and stand still
- **Every edit ends the walk**, beside `refused` and for the same reason: the walk is about one argument and a line that has changed is a different argument. What the last press put on the row stays there, which is what makes `tab` then a letter the ordinary way to take a candidate
- **Nothing is drawn while a walk is on the row.** What is there is a whole candidate already, and a second offer drawn against it would read as part of the same word
- **Grey, which is what the palette already spends on a value that is not one.** It is not in the line until `tab` puts it there, and the reader has to tell the two apart at a glance. It takes what the line leaves of the field rather than a field of its own, so a long command loses the offer instead of pushing `esc` and `enter` off the row. The terminal cursor sits on its first cell, which is where the cursor already was
- **The row does not name `tab`.** The prompt names its two mode keys and no more — `ctrl-w`, the arrows and home and end are unnamed there as well — and the suggestion under the cursor is its own affordance