The beginning of a parse-don't-validate-based hindley-milner
typechecker, which returns on success an IR where every AST node
trivially knows its own type, and using those types to determine LLVM
types in codegen.
We make use of the -O man=… option of mandoc(1) which allows to convert
cross references via the .Xr macro into actual hyperlinks in the output.
This can be disabled (by passing "none") or done in two modes:
* all: links all .Xr cross references as if they were in
$out/%N.%S.html. This will lead to broken links of course.
* inManDir: only link to files in $out if the man page is found in
manDir, use the template defined in linkXrFallback if not.
all is the default, since we don't require all man pages to be in
manDir, so it would be potentially confusing if the path attribute was
used in the pages list.
linkXrFallback uses the debian online man viewer by default currently,
since it can be decently hyperlinked and debian has a lot of packages.
Other options would be:
* https://manpages.ubuntu.com/manpages/latest/en/man%S/%N.%S.html
* https://man.archlinux.org/man/%N.%S.en
* https://man.openbsd.org/%N.%S
* https://www.man7.org/linux/man-pages/man%S/%N.%S.html
Change-Id: I1363b9dfdda25cb7383c7310b8115c335444bd3d
Reviewed-on: https://cl.tvl.fyi/c/depot/+/2597
Tested-by: BuildkiteCI
Reviewed-by: sterni <sternenseemann@systemli.org>
htmlman is a very simple nix based static site generator which is
intended for rendering HTML representations for man pages plus an index
page listing all available pages. For the sake of simplicity (and unlike
previous iterations of this piece of code) other documentation artifacts
and formats are not supported.
Usually web services like GitHub and depot's web interface are pretty
good at displaying "normal" documentation artifacts like markdown files,
but man pages are usually not rendered — with the additional problem
that it's source is virtually unreadable. htmlman should provide a
simple static site generator which can be plugged into GitHub actions or
the like to automatically generate rendered version of man pages tracked
in version control.
Change-Id: Ib53292964b3ff84c32d70c5fde257a2edb8c2122
Reviewed-on: https://cl.tvl.fyi/c/depot/+/2596
Tested-by: BuildkiteCI
Reviewed-by: tazjin <mail@tazj.in>
Reviewed-by: Profpatsch <mail@profpatsch.de>
Reviewed-by: sterni <sternenseemann@systemli.org>
The default for this seems to have changed in a recent notmuch
release.
Change-Id: I1542b20c2e3edf72a3472c5277bce313c6df12b8
Reviewed-on: https://cl.tvl.fyi/c/depot/+/2595
Reviewed-by: tazjin <mail@tazj.in>
Tested-by: BuildkiteCI
This also includes a fix for an issue where the identifiers of
variables were pushed onto the stack, which is incorrect.
Change-Id: Id89b388268efad295f29978d767aa4b33c4ded14
Reviewed-on: https://cl.tvl.fyi/c/depot/+/2594
Reviewed-by: tazjin <mail@tazj.in>
Tested-by: BuildkiteCI
identifier_str might look a bit overengineered, but we want to reuse
this bit of code and it needs a reference to the token from which to
pick the identifier.
The problem with this is that the token would be owned by self, but
the function needs to mutate (the interner), so this implementation is
the most straightforward way of acquiring and working with an
immutable reference to the token before interning the identifier.
Change-Id: I618ce8f789cb59b3a9c5b79a13111ea6d00b2424
Reviewed-on: https://cl.tvl.fyi/c/depot/+/2592
Reviewed-by: tazjin <mail@tazj.in>
Tested-by: BuildkiteCI
Making this function a macro instead makes it possible to match
arbitrary token kinds, even the ones that carry data, without changing
the syntax too much.
Change-Id: I5cda9e36d6833bd9c259f7d4d8340db6e783b4e8
Reviewed-on: https://cl.tvl.fyi/c/depot/+/2593
Reviewed-by: tazjin <mail@tazj.in>
Tested-by: BuildkiteCI
users.sterni.nix.utf8 implements UTF-8 decoding in pure nix. We
implement the decoding as a simple state machine which is fed one byte
at a time. Decoding whole strings is possible by subsequently calling
step. This is done in decode which uses builtins.foldl' to get around
recursion restrictions and a neat trick using builtins.deepSeq puck
showed me limiting the size of the thunks in a foldl' (which can also
cause a stack overflow).
This makes decoding arbitrarily large UTF-8 files into codepoints using
nix theoretically possible, but it is not really practical: Decoding a
36KB LaTeX file I had lying around takes ~160s on my laptop.
Change-Id: Iab8c973dac89074ec280b4880a7408e0b3d19bc7
Reviewed-on: https://cl.tvl.fyi/c/depot/+/2590
Tested-by: BuildkiteCI
Reviewed-by: sterni <sternenseemann@systemli.org>
switch would probably otherwise be called match, but has been renamed so
it isn't confused with string.match and the enum matching capabilities
yants has.
It implements the closest to pattern matching nix can come which is
still flexible enough to not be painful: Syntactically it works like
cond, but is given a value. Instead of booleans it checks passed
predicates or equality if simple values are passed. Both types of checks
can be mixed.
Change-Id: I40f000979cfd469316e15fd58d6c3a80312c1cc4
Reviewed-on: https://cl.tvl.fyi/c/depot/+/2589
Tested-by: BuildkiteCI
Reviewed-by: sterni <sternenseemann@systemli.org>
Since nix ends the substring at the end of the string anyways we can
just statically use the largest nix integer as the length of the string.
According to my testing this it ever so slightly faster as well.
Change-Id: I64566e91c7b223f03dcebe3bc5710696dc4261bc
Reviewed-on: https://cl.tvl.fyi/c/depot/+/2587
Tested-by: BuildkiteCI
Reviewed-by: sterni <sternenseemann@systemli.org>
After all it only matches strings.
Change-Id: I3d2e5221ef43f692de69028e78ed98b6b11f82d1
Reviewed-on: https://cl.tvl.fyi/c/depot/+/2586
Tested-by: BuildkiteCI
Reviewed-by: sterni <sternenseemann@systemli.org>
These aren't particularly useful without side effects, but one step at
a time.
This diverges slightly from the book, in that OpPop retains the last
value it "forgot" from the stack in a special field on the
interpreter.
This makes it possible to return values from expression statements,
which helps in cases where Lox is embedded as a scripting
language (please don't do this ever) or in tests.
Change-Id: Ided0bc04c6e80ddb23ba4693d61ac9e08b002d58
Reviewed-on: https://cl.tvl.fyi/c/depot/+/2584
Reviewed-by: tazjin <mail@tazj.in>
Tested-by: BuildkiteCI
This is again a step closer to the book, but there are some notable
differences:
* Only constants encountered by the compiler are interned, all other
string operations (well, concatenation) happen with heap objects.
* OpReturn will always ensure that a returned string value is newly
heap allocated and does not reference the interner.
Change-Id: If4f04309446e01b8ff2db51094e9710d465dbc50
Reviewed-on: https://cl.tvl.fyi/c/depot/+/2582
Reviewed-by: tazjin <mail@tazj.in>
Tested-by: BuildkiteCI
This is based on this matklad post:
https://matklad.github.io/2020/03/22/fast-simple-rust-interner.html
It's modified slightly to provide a safer interface and slightly more
readable implementation:
* interned string IDs are wrapped in a newtype that is not publicly
constructible
* unsafe block is reduced to only the small scope in which it is
needed
* lookup lifetime is pinned explicitly to make the intent clearer when
reading this code
Change-Id: Ia3dae988f33f8e5e7d8dc0c1a9216914a945b036
Reviewed-on: https://cl.tvl.fyi/c/depot/+/2578
Tested-by: BuildkiteCI
Reviewed-by: tazjin <mail@tazj.in>
... including concatenation.
This diverges significantly from the book, as I'm using std::String
instead of implementing the book's whole heap object management
system.
It's possible that Lox in Rust actually doesn't need a GC and the
ownership model works just fine.
Change-Id: I374a0461d627cfafc26b2b54bfefac8b7c574d00
Reviewed-on: https://cl.tvl.fyi/c/depot/+/2577
Tested-by: BuildkiteCI
Reviewed-by: tazjin <mail@tazj.in>
What you see here is mostly the fallout of me implementing a correct
urlencode implementation in nix for Profpatsch's blog implementation
(although they'll probably keep it at arm's length).
Where I want to go from here:
* Extend this library towards general purpose nix™, mainly by
implementing missing interfaces which you'd still have to use
<nixpkgs/lib> for right now. Reexposing parts of <nixpkgs/lib>
with better naming is fine for now, at some point I'd contemplate
making this depend on nothing outside of depot, maybe even itself
(should be easy we only use yants for an easily replaceable check).
* Improve error messages possibly by carefully reintroducing yants. I
originally typed essentially everything using yants, but turns out
this can a) be dangerous when stuff you are handling throws because
type checking means evaluating and b) has a incredible performance
cost in some cases.
* Reexpose builtins with better naming and slightly wrapped so they
don't unrecoverably throw in cases where a null or something would
suffice.
Change-Id: I33ab08ca4e62dbc16b86c66c653935686e6b0e79
Reviewed-on: https://cl.tvl.fyi/c/depot/+/2541
Reviewed-by: sterni <sternenseemann@systemli.org>
Reviewed-by: Profpatsch <mail@profpatsch.de>
Tested-by: BuildkiteCI
This makes it possible to specify the input & output types of the
binary_op macro. If only one type is specified, it is assumed that the
input and output types are the same.
Change-Id: Idfcc9ba462db3976b69379b6693d091e1a525a3b
Reviewed-on: https://cl.tvl.fyi/c/depot/+/2573
Reviewed-by: tazjin <mail@tazj.in>
Tested-by: BuildkiteCI
Adds support for true, false & nil. These each come with a new
separate opcode and are pushed directly on the stack.
Change-Id: I405b5b09496dcf99d514d3411c083e0834377167
Reviewed-on: https://cl.tvl.fyi/c/depot/+/2571
Reviewed-by: tazjin <mail@tazj.in>
Tested-by: BuildkiteCI
Introduces a new enum which represents the different types of possible
values, and modifies the rest of the existing code to wrap/unwrap
these enum variants correctly.
Notably in the vm module, a new macro has been introduced that makes
it possible to encode a type expectation and return a runtime error in
case of a type mismatch.
Change-Id: I325b5e31e395c62d8819ab2af6d398e1277333c0
Reviewed-on: https://cl.tvl.fyi/c/depot/+/2570
Reviewed-by: tazjin <mail@tazj.in>
Tested-by: BuildkiteCI
If I was adding any dependencies, this might be a good one for a
property-based test thing, but I'm not going to.
Change-Id: Ia801d041479d1a88c59ef9e0fe1460b3640382e3
Reviewed-on: https://cl.tvl.fyi/c/depot/+/2569
Reviewed-by: tazjin <mail@tazj.in>
Tested-by: BuildkiteCI
Without this fix we would keep parsing in the same precedence level
and get weird things like:
10 - -10 + 10
=> 10
Change-Id: If2bed4569fbf566027011037165a9b3c09b7427c
Reviewed-on: https://cl.tvl.fyi/c/depot/+/2567
Reviewed-by: tazjin <mail@tazj.in>
Tested-by: BuildkiteCI
This should clean up everything in the way of actually running this
end-to-end.
Change-Id: Ie89d82472a458256a251a4fddc1c36d88d21f5f2
Reviewed-on: https://cl.tvl.fyi/c/depot/+/2563
Reviewed-by: tazjin <mail@tazj.in>
Tested-by: BuildkiteCI
Defines a new precedence levels enum which can be used to restrict the
parser precedence in any given location. As an example, unary
expressions and grouping are implemented, as these have a different
precedence from e.g. expression()
Change-Id: I91f299fc77530f76c3aba717f638985428104ee5
Reviewed-on: https://cl.tvl.fyi/c/depot/+/2558
Reviewed-by: tazjin <mail@tazj.in>
Tested-by: BuildkiteCI
