hidden prep: Pass &mut GPlayerState to log_did_to_piece

Piece id conversion is going to possibly need to update the map in
gpl.

NFC.

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

daemon: wip: add a todo relating to PieceGone

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

hidden: wip a new Lens placeholder

(Actually, we changed our mind about this and it was never used.)

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

hidden: Declare new status of Lens

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

cargo: Update to our slotmap MR git branch

This does not seem to show up in the lockfile.  Odd.

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

hidden: Provide per-player id map

Not used yet...

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

hidden: wip per-player id map

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

Use local slotmap for VacantEntry::remove_stale_entry

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

hidden prep: Do not make PieceUpdateOps Clone

Tbis is not a good idea and not needed.

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

hidden prep: Allow api updates to be per-player

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

hidden prep: Allow prepared updates to be per-player

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

hidden prep: Add a few missing Clone/Copy

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

hidden prep: Per-player TransmitUpdateEntry can be None

NFC.

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

hidden prep: Pass PlayerId through TransmitUpdateEntry generation

NFC.

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

hidden prep: Pass PUO to piece_update_fallible

No functional change.

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

hidden prep: Reorganise PieceUpdateFromOp

We are going to want to allow an op to give different answers to
different clients.  The representation from an op is going to be
either basically what we have now, or a version where the
PieceUpdateOp is different for each player.

Prepare for this with some struct and type shuffling.  We laave a
convenience Into impl to allow ops to easily handle the simple case.

No functional change.

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

hidden prep: Pass player to json_len() methods

NFC right now.  We are going to want this later.

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

hidden: wip notes/plans, found

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

hidden: wip notes/plans

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

hidden: wip notes/plans, found

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

config: Allow OTTER_LOG to partially override config

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

utils: Provide toml_merge

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

config: Allow OTTER_LOG to (completely) override config

If the env var is supplied and results in a nonempty log spec, we
replace the log config completely.

This is quite a lot of faff because no-one likes to give us the right
interfaces.  In particular flexi_logger wants to give and receive
(byte) strings.

This is not the final state; we want to merge them, really.

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

toml-de: Provide from_slice

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

config: Refactor log setup into own block

We'll add stuff in a moment that modifies log while it's still a
`Table` rather than a `Value`.

NFC.

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

daemon: api route macros: tolerate attrs on structs

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

daemon: use macro for all api routes

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

daemon: wip macro for api routes

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

daemon: Fix many errors (OER rather than OE)

Firstly, all these functions shuuld throw OER which is the thing which
implements the Rocket Responder trait, so that the error actually goes
back to the client.

Secondly, they all need ?.  Without that, these functions all always
succeed and always return Result::Ok(some Result<, OE[R]>).
Surprisingly this is accepted, because Result<> impl Responder
and apparently Result<Result<..>> does too!

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

Remove a space

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

cargo: Update to lazy_init 0.5

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

daemon Cargo.toml: Fix dependency versions

We fixed this for src/ but daemon's Cargo.toml was resurrected from an
earlier state so missed it.

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

Makefile: pass cargo --workspace, (nearly) always

This seems to work around
  https://github.com/rust-lang/cargo/issues/9114
  https://github.com/rust-lang/cargo/issues/9114

The WASM call doesn't have it because (i) we want that to be minimal
(ii) there's only the one call to cargo for that architecture.

Build times:

  touch src/updates.rs; (time make -j12 wdt) 2>&1 |ts -s %.s
    before 26s    after 19s

  touch src/updates.rs; (time make -j12) 2>&1 |ts -s %.s
    before 33s    after 27s

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

Drop thirtyfour_sync from Cargo.nail, our MR was merged

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

cargo update

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

wdt: Apply test! to wdt-altergame

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

wdt: Provide ActionChainExt::move_pc for convenience

And use it in one place.  NFC.

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

wdt-simple: conflict: establish our preconditions

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

wdt: Tidy up test selection etc.

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

wdt: Allow test name argument

Currently only for wdt-simple, and not complete yet.

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

wdt: conflict test actually tests, and passes, yay!

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

utils: Improve ensure_eq! so it always takes references

That saves some minorly annoying & typing

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

wdt: wdt-simple: get client into gots

Not used yet

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

wdt: Provide WindowGuard::client()

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

wdt: wdt-simple: Remove an unused variable

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

wdt: remove some silly lifetime annotations nfc

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

wdt: import OnceCell into scope nfc

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

updates: Clean up an obsolete type alias

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

wdt: clean up conflict check

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

updates, errors: Completely revamp error handling

There were multiple bugs here.  We weren't coherent about what to send
to whom, and whether to update the piece generation, if there was an
error (including a conflict or a partially processed API op).

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

updates: Formatting fix (nfc)

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

updates: Break out some sub-functiosn

We're going to want to reuse this.

Broken out of a portmanteau commit.  NFC.

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

updates, errors: Add some missing Clone impls (nfc)

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

updates, errors: Break out PUE_Piece::json_len (nfc)

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

updates, errors: Reformat IsResponseToClientOp (nfc)

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

updates, errors: Add a doc comment

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

wdt: Use w.piece_held()

Rather than the clone-and-hack copy.  Tidying up, NFC.

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

wdt: conflict test, further wip

Not a complete test yet.

There are bugs to fix here.  In particular, right now, it seems that
both clients get a conflict ?

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

wdt: conflict test, further wip

Not a complete test yet.

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

wdt: wip test selection

This is very incomplete but it doesn't cause any *trouble* per se.

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

script: Add a log message for error piece update

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

wdt: drag off: synch before chk

Otherwise we can see the client's uncorrected view of things.

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

api: Add some extra debug for api op processing

This helps debug the conflict protocol (PROTOCOL.md).

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

utils: Improve display_as_debug! so it can pass generic params

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

introduce TransmitUpdateEntry_Piece (nfc)

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

introduce PreparedUpdateEntry_Piece (nfc)

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

wdt: Break out w.piece_held()

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

wdt: Fix unselect test to actually work

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

wdt: Tidy up new facilities

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

wdt: Use new facilities

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

wdt: new move_pos extension method

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

wdt: Use new w_move extension method

This necessitates using our branch of thirtyfour_sync until
  Make action_chain module pub
  https://github.com/stevepryde/thirtyfour_sync/pull/4

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

wdt: wip new pos handling trait

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

wdt: wip new pos handling trait

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

wdt: wip conflict test

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

wdt: replace an assert with an ensure

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

wdt: plumb server child through to Setup

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

wdt: drag_end test, works

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

wdt: wip drag_end test

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

wdt: nailing-cargo generate-lockfile, updates wasi

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

wdt: Drop direct toml dependency again

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

wdt: Actually read spec, and make it work

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

wdt: Provide game_spec_data

No caller yet.

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

wdt: Break out game_spec_path

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

wdt: Break out posg2posw, nfc

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

cargo update

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

updatee wasm-bindgen

Got this message:

  nailing-cargo: invoking: /home/ian/Rustup/Game/Build/wasm-pack/target/debug/wasm-pack --cargo-path=/bin/echo build --out-dir=../target/packed-wasm wasm -t no-modules --release
  \e[1m\e[2m[INFO]\e[0m: Checking for the Wasm target...
  \e[1m\e[2m[INFO]\e[0m: Compiling to Wasm...
  build --lib --release --target wasm32-unknown-unknown
  :-) \e[1m\e[2m[WARN]\e[0m: origin crate has no README
  \e[1m\e[2m[INFO]\e[0m: License key is set in Cargo.toml but no LICENSE file(s) were found; Please add the LICENSE file(s) to your project directory
  \e[1m\e[2m[INFO]\e[0m: Installing wasm-bindgen...
  error:

  it looks like the Rust project used to create this wasm file was linked against
  a different version of wasm-bindgen than this binary:

    rust wasm file: 0.2.69
       this binary: 0.2.70 (b6355c270)

  Currently the bindgen format is unstable enough that these two version must
  exactly match, so it's required that these two version are kept in sync by
  either updating the wasm-bindgen dependency or this binary. You should be able
  to update the wasm-bindgen dependency with:

      cargo update -p wasm-bindgen

  or you can update the binary with

      cargo install -f wasm-bindgen-cli

  if this warning fails to go away though and you're not sure what to do feel free
  to open an issue at https://github.com/rustwasm/wasm-bindgen/issues!

  Error: Running the wasm-bindgen CLI
  Caused by: failed to execute `wasm-bindgen`: exited with exit code: 1
    full command: "/home/rustcargo/.cache/.wasm-pack/wasm-bindgen-454f7c61ea4f4e72/wasm-bindgen" "/volatile/rustcargo/Rustup/Game/server/target/wasm32-unknown-unknown/release/otter_wasm.wasm" "--out-dir" "wasm/../target/packed-wasm" "--typescript" "--target" "no-modules"
  nailing-cargo: really failed (exit status 256)
  nailing-cargo: unnailed.  status 1.
  make: *** [Makefile:199: stamp/wasm-pack] Error 1

  Compilation exited abnormally with code 2 at Tue Jan 26 22:35:13

This message appeared after a `make clean` and then rerunning the
build "fixed" it; `make clean` produced it again.  This is not
confidence-inspiring.

cargo update fixed it.

But I thought wasm-pack was supposed to handle this properly.  Not
sure why I am using it, now.

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

Add a todo note about wasm ABI

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

remove redundant #![feature...]

Updated nightly, now this is stable, yay!

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

remove redundant semicolons

Updated nightly, new warning.

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

whitespace changes

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

Postpone a couple of JS xxx into todo

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

Change Coord to i32.

Size should not depend on our pointer size, so isize is wrong.

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

cargo update

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

make clean: Remove emacs backup files too

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

Remove three obsolete files

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

bump version to 0.3.0

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

library: Adjust red to be a little less pale

This makes the red pieces less pink-looking.

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

library: Make dark pieces (purple and red) have pale edges

This makes them easier to spot.

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

infopane: styling

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>