Our cable has a rather complex getCollisionBoundingBox implementation,
which (combined with computing the actual block state) ends up taking an
awful lot of time.
This shaves 9% off each frame. I don't miss 1.12 one bit.
Previously we would compute the current timeout flags every 128
instructions of the Lua machine. While computing these flags is
_relatively_ cheap (just get the current time), it still all adds up.
Instead we now set the timeout flags from the computer monitor/watchdog
thread. This does mean that the monitor thread needs to wake up more
often[^1] _if the queue is full_, otherwise we can sleep for 100ms as
before.
This does mean that pausing is a little less accurate (can technically
take up 2*period instead). This isn't great, but in practice it seems
fine - I've not noticed any playability issues.
This offers a small (but measurable!) boost to computer performance.
[^1]: We currently sleep for scaledPeriod, but we could choose to do less.
When calling .flip(), we limit the size of the buffer. However, this
limit is not reset when writing the next time, which means we get
out-of-bounds errors, even if the buffer is /technically/ big enough.
Clearing the buffer before drawing (rather than just resetting the
position) is enough to fix this.
Fixes#476 (and closes#477, which is a duplicate)
- Use texture over texture2D - the latter was deprecated in GLSL 1.30.
- Cache the tbo buffer - this saves an allocation when monitors update.
Closes#455. While the rest of the PR has some nice changes, it
performs signlificantly worse on my system.
This moves monitor networking into its own packet, rather than serialising
using NBT. This allows us to be more flexible with how monitors are
serialised.
We now compress terminal data using gzip. This reduces the packet size
of a max-sized-monitor from ~25kb to as little as 100b.
On my test set of images (what I would consider to be the extreme end of
the "reasonable" case), we have packets from 1.4kb bytes up to 12kb,
with a mean of 6kb. Even in the worst case, this is a 2x reduction in
packet size.
While this is a fantastic win for the common case, it is not abuse-proof.
One can create a terminal with high entropy (and so uncompressible). This
will still be close to the original packet size.
In order to prevent any other abuse, we also limit the amount of monitor
data a client can possibly receive to 1MB (configurable).
- cc.pretty.pretty now accepts two additional options:
- function_args: Show function arguments
- function_source: Show where functions are defined.
- Expose the two options as lua.* settings (defaulting function_args to
true, and function_source to false).
These are then used in the Lua REPL.
Closes#361
- Use jacoco for Java-side coverage. Our Java coverage is /terrible
(~10%), as we only really test the core libraries. Still a good thing
to track for regressions though.
- mcfly now tracks Lua side coverage. This works in several stages:
- Replace loadfile to include the whole path
- Add a debug hook which just tracks filename->(lines->count). This
is then submitted to the Java test runner.
- On test completion, we emit a luacov.report.out file.
As the debug hook is inserted by mcfly, this does not include any
computer startup (such as loading apis, or the root of bios.lua),
despite they're executed.
This would be possible to do (for instance, inject a custom header
into bios.lua). However, we're not actually testing any of the
behaviour of startup (aside from "does it not crash"), so I'm not
sure whether to include it or not. Something I'll most likely
re-evaluate.
`local varname = value` results in `varname` being inaccessible in
the next REPL input. This is often unintended and can lead to confusing
behaviour. We produce a warning when this occurs.
This uses the system described in #409, to render monitors in a more
efficient manner.
Each monitor is backed by a texture buffer object (TBO) which contains
a relatively compact encoding of the terminal state. This is then
rendered using a shader, which consumes the TBO and uses it to index
into main font texture.
As we're transmitting significantly less data to the GPU (only 3 bytes
per character), this effectively reduces any update lag to 0. FPS appears
to be up by a small fraction (10-15fps on my machine, to ~110), possibly
as we're now only drawing a single quad (though doing much more work in
the shader).
On my laptop, with its Intel integrated graphics card, I'm able to draw
120 full-sized monitors (with an effective resolution of 3972 x 2330) at
a consistent 60fps. Updates still cause a slight spike, but we always
remain above 30fps - a significant improvement over VBOs, where updates
would go off the chart.
Many thanks to @Lignum and @Lemmmy for devising this scheme, and helping
test and review it! ♥
Unknown blit colours, such as " " will be translated to black for the
background or white for the foreground. This restores the behaviour from
before #412.
- Write to a PacketBuffer instead of generating an NBT tag. This is
then converted to an NBT byte array when we send across the network.
- Pack background/foreground colours into a single byte.
This derives from some work I did back in 2017, and some of the changes
made/planned in #409. However, this patch does not change how terminals
are represented, it simply makes the transfer more compact.
This makes the patch incredibly small (100 lines!), but also limited in
what improvements it can make compared with #409. We send 26626 bytes
for a full-sized monitor. While a 2x improvement over the previous 58558
bytes, there's a lot of room for improvement.
This functions the same as shell.run, but does not tokenise the
arguments. This allows us to pass command line arguments through to
another program without having to re-quote them.
Closes#417
- Remove the parenthesis around the text (so it's now
"Computer ID: 12"), rather than "(Computer ID: 12").
- Show the tooltip if the computer has an ID and no label (as well as
when in advanced mode).
- Lint references to unknown fields of modules, excluding the keys and
colours modules. This caught several silly errors in our stub files,
but nothing else.
- Lint on using unknown globals. This highlighted a couple of really
silly mistakes. Fixes#427.
- Add documentation for fs.attributes, fs.getCapacity and pocket, as
they were not defined before.
Co-authored-by: JackMacWindows <jackmacwindowslinux@gmail.com>
- I'm excluding pocket computers, as they have such a tiny screen I'm
not sure the screen estate is worth it.
Pocket computers /generally/ aren't people's starter machine, so I
think this is fine.
- Prune the motd list, and try to make them a little shorter. I think
this list is more of the interesting ones. We can modify this list in
the future, as we get more feedback.[^1]
- Also fix paint/edit not adding an extension when they should. This
was caused by the settings rewrite, as the explicitly provided
default shadowed the one provided by bios.lua.
[^1]: ~5 months ago I asked for some feedback about enabling motds by
default. I only got something constructive back today >_>.
- The store is now split into two sections:
- A list of possible options, with some metadata about them.
- A list of values which have been changed.
- settings.define can be used to register a new option. We have
migrated all existing options over to use it. This can be used to
define a default value, description, and a type the setting must have
(such as `string` or `boolean).
- settings.{set,unset,clear,load,store} operate using this value list.
This means that only values which have been changed are stored to
disk.
Furthermore, clearing/unsetting will reset to the /default/ value,
rather than removing entirely.
- The set program will now display descriptions.
- settings.{load,save} now default to `.settings` if no path is given.
This is a backport of 1.15's terminal rendering code with some further
improvements. This duplicates a fair bit of code, and is much more
efficient.
I expect the work done in #409 will supersede this, but that's unlikely
to make its way into the next release so it's worth getting this in for
now.
- Refactor a lot of common terminal code into
`FixedWithFontRenderer`. This shouldn't change any behaviour, but
makes a lot of our terminal renderers (printed pages, terminals,
monitors) a lot cleaner.
- Terminal rendering is done using a single mode/vertex format. Rather
than drawing an untextured quad for the background colours, we use an
entirely white piece of the terminal font. This allows us to batch
draws together more elegantly.
- Some minor optimisations:
- Skip rendering `"\0"` and `" "` characters. These characters occur
pretty often, especially on blank monitors and, as the font is empty
here, it is safe to skip them.
- Batch together adjacent background cells of the same colour. Again,
most terminals will have large runs of the same colour, so this is a
worthwhile optimisation.
These optimisations do mean that terminal performance is no longer
consistent as "noisy" terminals will have worse performance. This is
annoying, but still worthwhile.
- Switch monitor rendering over to use VBOs.
We also add a config option to switch between rendering backends. By
default we'll choose the best one compatible with your GPU, but there
is a config option to switch between VBOS (reasonable performance) and
display lists (bad).
When benchmarking 30 full-sized monitors rendering a static image, this
improves my FPS[^1] from 7 to 95. This is obviously an extreme case -
monitor updates are still slow, and so more frequently updating screens
will still be less than stellar.
[^1]: My graphics card is an Intel HD Graphics 520. Obviously numbers
will vary.
Jonathan Coates