die: Hack about at SVG to make a nicer template for label

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

dice: Add labels to demo game

But they don't display yet...

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

dice: Add cross-referencing notes and comments

This may help us find things in future.  Used inkscape's "document
properties", "metadata" pane, and then manually moved the new metadata
tag nearer to the top of the svg file.

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

dice: Fix cooldown radius

This looks good now.

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

Provide cooldown timer stroke width to Rust

We now have this in two places which is a bit poor, but we already
have some duplication between die.svg and the Rust code - eg, the path
generation which was just c&p.

I have left a comment.  And arranged to filter it out for distribution.

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

Provide select stroke width to Rust

We now have this in two places which is a bit poor, but I don't fancy
plumbing it into the ts machinery.

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

dice: Use bespoke outline path for surround (selection)

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

OutlineTrait: Document outline_path

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

dice: Fix a radius/diameter confusion

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

die svg: Shrinking by 10: Change window

Just asked inkscape to recenter the view.

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

die svg: Shrinking by 10: Reduce die box size

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

die svg: Shrinking by 10: Reduce stroke widths

We don't have anything to rescale the elements we extract for use as
the cooldown timer, so we need to get the linewidths as we finally
want them.

In this commit we mostly do the stroke widths and the circle size.

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

dice: Adjust sizes, wip

Adjust the strokes of the circles using inkscape.

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

dice: Add a test die to the demo game

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

dice: Cooldown template: Add header about autogeneration

This may help avoid editing it manuallly...

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

dice: Make moveable

Don't bake into the piece implementation that these aren't moveable.
If this is desired, it can be part of the game spec.

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

dice: Sort out defaults

 * Abolish the separate DEFAULT_COOLDOWN constant.  This seems otiose.
 * Clarify what the radius scale is supposed to mean
 * Provide a default radius scale

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

test.table.toml: Add a currently-working rune

Many of the others are obsolete but I don't want to take the time now
to clean them up.

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

dice: Start a cooldown the first time a die is added to the game

This is useful for testing.  It also means you can't just add a die
showing what you want using the game management interface...

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

Provide InternalError::FutureInstantOutOfRange

The dice itself wants to sometimes do this conversion (notably, on
initial piece load), and if it's out of range that's implausible.

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

dice: Always use a circular Outline and adjust radiii

The circular outline will go nicely with the cooldown timer.

The radius etc. is still not quite right.

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

dice: Set cooldown_expiry to None when we can

This will avoid pointless calls to Instant::now, which would otherwise
be needed to discover that the cooldown time has expired (perhaps long
ago).

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

dice: Tidy remprop handling

This macro was overkill.  It existed mostly to avoid uselessly
comparing the remaining duration with 0 and storing None when it was
there already.  That's a daft microoptimisation.

Doing it this way instead lets us only do the remprop calculation when
it's wanted, and it will let us provide a way to set cooldown_expires
to None if and when we have &mut State and can see it's timed out.

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

dice: Initial implementation of piece

Does not work properly yet.

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

Provide FutureTimeInstant, for use by dice

This loads and saves as a Duration.

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

dice: cooldown_path etc.: rename remaining proportion to "remprop"

This distinguishes "remaining" (a duration) from "remprop" (a proportion).

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

dice: Cooldown template: Use total_ms for duration variable

Make the units clearer.  We will provide a value for this from a
Serialize context struct in the Rust code, which is still to come.

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

dice: Generate cooldown template using extractor

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

Move die helper materials to dice/ subdirectory

This declutters the toplevel a bit.

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

Special client rendering: Do explain why not to provide xdata here

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

Special client rendering: Fix quoting of JSON payload

The needs to be in " " (and the JS will evaluate it).

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

Provide plumbing for special client rendering instructions

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

Make PieceBaseTrait dyn_upcast-able

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

pcrender: prep for more attributes in pelem

We are going to add a .special.

NFC.

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

die: wip svg, add more tests

Again, verified by hand-hacking and using Refresh in inkscape, that
they look right.

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

die: wip svg, cooldown timer better assert

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

die: wip svg, fix cooldown timer path and add test cases

I've verified that the 1. test case shows as expected.

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

die: wip svg, make example less than half

This let me c&p the string into the new test case.  Although, the test
will need to be different because of rounding errors.

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

dice: Provide die_cooldown_path

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

die: wip svg, fix colour of white

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

die: wip svg extractor

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

die: wip svg, change timewhite path (back) to a circle

We want to be able to auto-adjust it by fiddling with r=.  That's
going to be awkward with the path element but easy with circle.

I have checked that one can (eg) change the width of the line in
inkscaqpe, and it remains a circle.

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

die: wip svg, remove unneeded parts

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

die: wip svg, remove zigzag

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

die: wip svg

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

die: wip svg

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

die: wip svg

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

InertPieceTrait: Add comment clarifying FaceId etc.

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

Gamestate: Fix a more style nit

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

SpecError: Add a comma (for consistent coding style)

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

Provide load_inert for SimplePieceSpec

This will let us use these as backgrounds for dice.

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

Move fn load out of SimplePieceSpec into macro

The only call site was in impl_PieceSpec_for_SimplePieceSpec.
Hoisting it there makes little codegen difference but seems more
natural.

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

Combine impls into impl_PieceSpec_for_SimplePieceSpec

We want to add load_inert to this.  It's getting too repetitive.

We can't use generic impl on <P: SimplePieceSpec> because we need to
specxify the serde typetag.

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

Pass face to InertPieceTrait::describe_html

This ought to have been part of
  Extend InertPieceTrait to support multiple faces

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

Make OccultIlkName a type alias, not a newtype

There's not really an abstraction here defending anything, and it
lacks constructors etc.  Just abolish it.

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

Plumb xdata through to InertPieceTrait svg

Occulted dice will need this to display the cooldown timer.

This makes dice's InertPieceTrait impl not very inert, but it is OK
because nothing can set off an inert die's cooldown timer.

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

Move itemname into PieceBaseTrait

We want this becasue dice want to use inert pieces as images, and use
their images' item names in their own item names.

This involves providing an itemname for ItemInertForOcculted.
Happily the SvgBaseName will do nicely.

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

Introduce PieceBaseTrait and put nfaces() in it

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

humantime: Switch to humantime_serde

We are going to want this for dice cooldown spec.
Using the humantime re-export avoids version problems.

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

humantime: Promote dependency to otter main crate

We are going to want this for dice cooldown spec.

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

spec errors: Expand WrongNumberOfFaces error

We're going to want to reuse this in a new context.  And, this already
produces better error messages for the case where it's used now.

It seems better to provide this rather generic error rather than
expect each funky piece kind to invent its own kind in SpecError.

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

docs: Document scope of semver guarantees

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

Make GenericSimpleShape::count_faces return RawFaceId

RawFaceId is smaller than usize.  It is better not to ask callers to
do the unwrap.  Motivated by wanting add a new call site which wants a
RawFaceId.

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

Extend load_inert to offer an occultable piece

Now inert pieces can be occultable too.  Currently this is meaningless
since we only call load_inert when resolving the occultation
for *another* piece, where we discard the occultation information.

But that is going to change.

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

Genericise SpecLoaded

We are going to want to make inert pieces occultable too.  Currently
this is meaningless since we only call load_inert when resolving the
occultation for *another* piece.  But that is going to change.

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

Fix message for IE::JSONEncode

The name is JSONEncode, so this must be serialisation, and indeed that
is true of both the construction sites.

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

Implement InertPieceTrait for SimpleShape

Right now, this allows simple circles and rectangles to be used as
backs (for occultation).  It will also allow them to be used for dice,
when we have those.

It is not convenient to provide a test for this because occultation
methods must be provided in libraries, and we don't have a test
library.

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

Style: tidy up and use fehler

Empirically, without the let r = ...; r, fehler and if_chain! produce
a type error.  That must be why we had the fehler use commented out.
But this way it works.

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

Extend InertPieceTrait to support multiple faces

We want this because we want to use InertPiece for dice and a die can
have multiple faces, which we want to specify as a single piece.  Ie,
the die face images will be an inert version of the die.

No functional change right now - we just move the default() call out
from the trait impls to each of the call sites.  In the shapelib
impl (which is the only impl right now) we pass through the face to
the rendering code (which lives in the main PieceTrait impl).

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

Introduce "inert" terminology for images for occultation

This is much less confusing.  In particular, I'm trying to distinguish
the occulted piece from the piece (or trait object) which the occulted
piece provides (to the caller of `load`) for its occulted image.

No functional change other than wording in an error message.

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

Replace a Mutex which was being an open-coded Lazy

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

Rename make_svg_defs

This is the thing that generates SVG.  That's worth mentioning.

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

Remove loaded_via_alias

This effectively reverts 707b43736d838436777f53b230e64054ecb244c5.

I have checked that this field is not used anywhere, including in any
relevant serialised output eg via the command listener.  As for game
save compatibility, we can simply ignore it on read (and, even go
backwards since it's Option).

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

Cargo.toml: Update all versions to 1.0.0

Commit made automatically by update-versions.

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

changelog: Document that this is version 1.0.0

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

changelog: Document changes since 0.7.3

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

daemon: Move src_service into services! call

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

inventory: Postpone this TODO

So far things still seem to be working, despite upstream rust problems
with #[used].

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

Document new build disk space requirement - reduced !

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

daemon: Reorder routes in order of likely frequency

I'm not sure if this is at all relevant but it seems sensible
and is at worst harmless.

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

actix: Revamp /_/{resource} handling

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

release checklist: Drop entry about nightly

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

Formulaic doc comments for other crates

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

Bespoke doc comments for otter, otter-cli and otter-daemon

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

Drop unused dependency: failure

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

cargo: Update zip (zipfile) dependency

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

Cargo update

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

Makefile: Build wasm-bindgen with selected rust version too

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

Deploy release build

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

Turn on integer overflow checks for release builds

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

actix: Fix CORS

The docs aren't entirely clear but suggest that the default allowed
origins is All.  But dbg! output seems to contradict this.  And we
should support more headers, and say that we don't vary our CORS at
all.

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

Docs update: Not beta, works on stable

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

Makefile: Arrangements for honouring RUST_VERSION

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

Update TODOs

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

clippy: Miscellaneous minor changes

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

clippy: Miscellaneous minor changes, and allow lints

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

clippy: Miscellaneous minor changes

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

clippy: Miscellaneous minor changes, and allow lints

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

clippy: Miscellaneous minor changes

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

apitest: Fix a write() call that should be write_all()

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