Break out ColourSpecExt::resolve

Hrm, maybe this should have been load() ?  This will do for now.

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

demo.game.toml: Remove non-hnoured min_unit setting

We decided to leave this to the players.

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

Html: Fix Debug truncation to not randomly panic

Ow!

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

spec loop detecton: Write down the rule

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

spec loop detecton: Detect for magic too.

Introduce PieceLoadArgs::recursing, and use it in the one current call
site.

Make the error more general.  So now we can call it for magic too.

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

clippy: Misc minor improvements

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

clippy: Remove a redundant pair of parens

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

cli bundles: autozip: Use Deflate

This is considerably faster.

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

progress reporting: Nest: Provide and use more desc plumbing

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

cli bundle progress: Say we're hashing, when we are.

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

cli bundle progress: Report for files in the zip we're building

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

cli bundle progress: Collect local zipfile members in a vec

This will let us easily count them, which we want to do for progress
reporting.

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

progress: Nest: Fix to actually be progressive

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

cli bundle progress: clear before thinking about bundles

Otherwise the output is a bit messed up.

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

progress: Nest: Issue a report at outer phase transitions

There's not much to say here, but without this, if the inner phase
does nothing, the user sees nothing.  Better to do something here.

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

otter cli: Use new Nest progress during bundle uploading

This lets us show progress of zipfile preparation.

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

progress: Replace Nest

This version doesn't need everyone to have counts.  In particular, you
can have woolly phases.  The otterlib bundle processing will want
this.

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

progress: Rename NestEqual from Nest

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

progress: Allow ProgressInfo's Count to contain only a proportion

We're going to introduce a thing that will use this.

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

bundle auto-zip: works, but needs some progress reporting

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

bundle auto-zip: wip

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

dice: Set label colour of red dice.  Hooray!

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

dice: Allow setting label colour

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

dice: Use new templating arrangements to do away with d8 group

Now we can have different dice all ni the smme group.

They all have to have the same size, so we adjust the d8 svg
accordingly.  (The d6 svg has to be the way it is because it's
extracted from the die cooldown etc. template.)

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

bundles: Fix a bug which could cause IE on bundle load

We need to make an entry entry in the bundles notes array even if the
hash is missing.  Or we might try to reuse the bundle number, if there
was a failed bundle upload followed by a server restart.

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

shapelib bundle tests: Do not filter out "filenames" after ^:

These are actually extra field definitions.  If there are these, our
parser is too stupid.  We're just filtering out the source filenames
to know that otter isn't reading them.

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

shapelib subst: Reorder other settings

This lets the explicit substs from various places override the builtin
image and colour.

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

shapelib subst: Reorder overrides

This is a bit confusing.  Early settings take precedence since they
get substituted away!

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

shapelib, subst: Allow colours to provide magic substs

Eg, for text colour.  The dice want this.

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

shapelib, subst: Allow groups to provide magic substs

Otherwise they might have to recapitulate the template.

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

shapelib subst: Prepare for substituting more into magic

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

docs: Document die numeric labels

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

run-inkscape-extension: Work around a weird inkscape behaviour

Without this, generic recolour instructions in shape libraries don't
work right if some of the recolourings don't apply to some of the
image files in the group.  Which is about to be the case.

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

subst: Use ${...} format internally for most things

This saves allocating in the usual case and is also clearer.

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

subst tests: Add test cases for converting needles

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

subst: Have substitutor handle any needle syntax

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

subst: Have subst_general return the resolved needle

This corrects the error reporting to show what we actually looked for.

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

subst test: Add many dbg calls

It turns out that I can make these fail and having it debug is useful.

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

dice: Change the inner edges in the blue d8

This looks better.  We will have to recolour them for the red die.

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

subst: Substitute x* variables into magic template

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

dice: red wip

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

dice: d8 wip

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

subst: Allow non-'static keys

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

magic: Use new ${image} syntax

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

subst test: Test format 2 more thoroughly

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

subst tests: Disable under miri

We're going to add dbg! calls, which break.

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

subst test: Test format 2

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

subst test: Test mf1 Filename

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

subst test: Reintroduce and fix up for new API

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

subst: Make some things into methods

Wanted to do this for subst and substn but this terminally confuses
the borrow checker.

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

subst: Always substitute dollars in mformat 2 text

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

subst: Provide Dollars::enabled

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

subst: Hoist dollars into c_* aliases

So now colour_subst_1 knows if it's text or not.
No functional change.

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

subst: Demote PerhapsSubst

Reverts
  subst: Promote PerhapsSubst

We shouldn't do this for anything but colour specs.

And, really, we ought to do it only for the item name and not for the
other things, or the ${$} de-escaping is very confusing.

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

subst: Track whether we're working in a filename

Ie, we are going to do this in a principled way, depending on what
kind of thing we are trying to generate, rather than just whether we
saw something other than "_c".

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

subst: Provide and use internal_err()

We're going to have more things that might be wrong at runtime...

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

subst: Break out Substituting::err

We're going to want to reuse this.

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

Make InternalLogicError be Clone, with an Arc

shapelib wants this for SubstErrors which ought to be Clone.

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

subst: Implement format 2 and convert all libraries

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

subst: Have internal functions take borrows

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

subst: Plumb mformat into Substituting, and change it to be Cow

This will allow us to do know whether to use the new format.

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

subst: Temporarily comment out the test

There is so much reorg here.  I don't want to keep editing this.

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

subst: Track in type whether we're substituting

This stops us from failing to do a final cleanup step (which is going
to be replacing `${$}` when we have mformat 2 style.  It also stop us
restarting substitution unexpectedly, and marks places where text
enters the substitution machinery.

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

subst: Promote PerhapsSubst

I thought we wanted this for sorts.  I find we don't, but I don't want
to drop this change and deal with the conflicts.  I will revert it
later.  Much later.

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

subst: Make PerhapsSubst::finish fallible

It is going to throw SubstError.

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

subst: Introduce PerhapsSubst which will let us have subst chaining

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

dice: d8 wip

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

dice: d8 wip

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

dice: d8 wip

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

otterlib: Show large objects a bit bigger

This algorithm is a bit paants.

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

otterlib: Show three views only for first two faces

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

otterlib: Fix column handling in preview

This was very confused.

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

dice: Get size right in preview with a special entrypoint

This is the code from
  dice: Get size right in preview by setting bbox, breaking other things
but in a new entrypoint used only for preview.  The comment explains why.

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

Revert "dice: Get size right in preview by setting bbox, breaking other things"

This reverts commit 1df93eed42dc37563a7bf5df825f800cb871396d.

dice: Get size right in preview by setting bbox, breaking other things

This is wrong.

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

dice: Unwrap surround_path

We're making changes here and it's confusing.  Make this more
prominent.

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

dice: Delegate our shape and bbox to outline

The bbox is still wrong.

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

dice: Preview partially-cooled-down

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

specs: Adjust shapes to be right with new code

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

dice: Adjust circle size to be right for images which are circles

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

Add shape method to OutlineTrait

dice will use this to decide their own outline.

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

impl OutlineTrait via ambassador for GenericSimpleShape

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

impl_via_ambassador: Support generics

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

impl OutlineTrait via ambassador for many types

We're going to add a method...

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

Move shapelib Outline structs into outline

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

Move outline into its own file

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

Unify outlines from shapelib and spec

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

Macro-ise spec::outline

We're going to do more here...

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

Rename RectOutline and CircleOutline

These are outlines and we are going to want an enum for just the
shapes.  Well, that's the Defn from shapelib but we want it more
generally.

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

shapelib: Rationalise some imports

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

otterlib: Show selection ring *over* the piece

This is what the game UI does.

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

piece API: rename save_reloaded_hook

This is clearer, given that we have load() for turning a spec into a
loaded piece.

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

support: Move slotmap

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

support: Move digestrw and timedfd

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

support: Move config

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

support: Rename PacketFrame errors

That's what they are.

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

support: Move packetframe et al.

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

support: Move childio and debugmutex

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

support: Move tz file to right directory

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

crate structure: Write down the import structure doctrine.

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