| Age | Commit message (Collapse) | Author | Files | Lines |
|---|
|
hpack is a bit dumb when generating the list of modules for a cabal
file's component if multiple of them live in the same directory.
Specifically it seems to assume that all modules in the source-dirs
of a particular component are also necessary for its compilation.
This is quite bad in the case of xanthous since both library and
executable have source-dirs: src, so all modules will be compiled
twice: Once for the library and then again for the executable
despite it depending on the library (actually 4 times in total
since we need to build a unprofiled and profiled object for each
module…).
To fix this we just move Main.hs into its own directory and change
the executable's source-dirs, so hpack doesn't get confused anymore.
Since all components now have their own source-dirs, unnecessary
redundant compilation should be down to 0. The diff of the cabal
file shows quite nicely how many module recompilation we've gotten
rid of.
Change-Id: I2df4fab9b0299b3a2b5d3005508c79b2d9796039
Reviewed-on: https://cl.tvl.fyi/c/depot/+/3533
Tested-by: BuildkiteCI
Reviewed-by: tazjin <mail@tazj.in>
Reviewed-by: grfn <grfn@gws.fyi>
|
|
Previously, we were using `smallestNotIn` for selecting new characters
for menu items with duplicate chatacters - this uses the 'Bounded'
instance for the type, which for Char meant the first character we would
always select was \NUL - making it look like the menu item had no
character, and making it impossible to actually select the menu item.
This introduces an AlphaChar newtype, which is a wrapper around Char
whose Bounded and Enum instances only use alphabetic characters (a-ZA-Z)
and uses that for menu characters instead.
Change-Id: If34ed9e9ce84f2bcb1cb87432cc6273f40b69f72
Reviewed-on: https://cl.tvl.fyi/c/depot/+/3229
Reviewed-by: grfn <grfn@gws.fyi>
Tested-by: BuildkiteCI
|
|
Allow the itemType raw to have density and volume fields, both of which
represent *intervals* of both density and volume (because both can
hypothetically vary a bit). The idea here is that when we're making
an *instance* of one of these items, we pick a random value in the
range.
Lots of stuff in this commit is datatype and typeclass instances to
support things like intervals being fields on datatypes that get
serialized to saved games - including a manual definition of Ord for
Item since Ord isn't well-defined for intervals
Change-Id: Ia088f2f75cdce9d00560297e5c269e3310b85bc3
Reviewed-on: https://cl.tvl.fyi/c/depot/+/3225
Reviewed-by: grfn <grfn@gws.fyi>
Tested-by: BuildkiteCI
|
|
2000 ticks after the character damages their fists by hitting something,
the character now develops calluses on their fists (scaled by *how*
damaged they've become) that reduce the chance of them receiving
additional damage from hitting things - up to a max of 5, which
prevents *all* damage from fistfighting.
This is all tracked in a new "Knuckles" struct in a new "Body" struct on
the character datatype, which manages stepping itself forward as part of
the Brain impl on the character.
Change-Id: Ica269f16fb340fb25900d2c77fbad32f10c00be2
Reviewed-on: https://cl.tvl.fyi/c/depot/+/3222
Reviewed-by: grfn <grfn@gws.fyi>
Tested-by: BuildkiteCI
|
|
I didn't end up using this directly for the thing I was doing, but it
still seems generally useful enough that I'm keeping it around for now
Change-Id: I05c8902d75845f2230ec2373a9677d61cfaafafd
Reviewed-on: https://cl.tvl.fyi/c/depot/+/3206
Reviewed-by: grfn <grfn@gws.fyi>
Tested-by: BuildkiteCI
|
|
Add a new "speech" generator module, with the beginnings of the vague
definition of the phonotactics of the language (there's one in here for
English based on the wikipedia article for English phonology, but it's
less than ideal as it has generated words like "sprurlkt") and the
ability to generate random syllables of a language by picking an onset,
nucleus, and coda from the list for that language (within a range of the
number of allowed of each syllable part). This will be used down the
road to automatically generate utterances from various
non-english-speaking creatures (so the accuracy is less important, just
that it "feels real").
Change-Id: I7b81375ec595239c05c5c800cbde1a2a900e38ac
Reviewed-on: https://cl.tvl.fyi/c/depot/+/3202
Reviewed-by: grfn <grfn@gws.fyi>
Tested-by: BuildkiteCI
|
|
I'm going to start adding generators for things like text soon, so it
makes sense to specifically sequester level generators as their own
thing
Change-Id: I175025375204fab7d75eba67dd06dab9bd2939d3
Reviewed-on: https://cl.tvl.fyi/c/depot/+/3201
Reviewed-by: grfn <grfn@gws.fyi>
Tested-by: BuildkiteCI
|
|
Memoize the return value of characterVisiblePositions to a new,
semi-abstracted "memo" field on the GameState, recalcuclated if the
character position ever changes. I'm 90% sure that the perf issues we
were encountering were actually caused by characterVisiblePositions
getting called once for *every tile* on draw, but this slightly larger
change also makes the game perform relatively-usably again.
Since this is only recalculated if the character position changes, if we
ever get non-transparent entities moving around without the characters
influence (maybe something building or knocking down walls?) we'll have
an issue there where the vision won't be updated as a result of those
changes if they happen while the character is taking a non-moving action
- but we can cross that bridge when we come to it.
Change-Id: I3fc745ddf0014d6f164f735ad7e5080da779b92a
Reviewed-on: https://cl.tvl.fyi/c/depot/+/3185
Reviewed-by: grfn <grfn@gws.fyi>
Tested-by: BuildkiteCI
|
|
Rename my //users directory and all places that refer to glittershark to
grfn, including nix references and documentation.
This may require some extra attention inside of gerrit's database after
it lands to allow me to actually push things.
Change-Id: I4728b7ec2c60024392c1c1fa6e0d4a59b3e266fa
Reviewed-on: https://cl.tvl.fyi/c/depot/+/2933
Tested-by: BuildkiteCI
Reviewed-by: tazjin <mail@tazj.in>
Reviewed-by: lukegb <lukegb@tvl.fyi>
Reviewed-by: glittershark <grfn@gws.fyi>
|
