commit_hash
stringlengths
40
40
author
stringlengths
1
57
date
timestamp[s]date
2010-07-26 04:45:09
2026-04-14 18:21:10
message
stringlengths
8
1.39M
diff
stringlengths
68
51.2k
files_changed
int64
1
136
insertions
int64
0
2.35k
deletions
int64
0
1.9k
1724addc4e1aa019a5ef068ba38b30db70e936e0
Kyle Mayes
2016-05-16T19:05:04
Reduce required Clang static libraries
diff --git a/build.rs b/build.rs index 051ab37..800ceb4 100644 --- a/build.rs +++ b/build.rs @@ -94,26 +94,18 @@ fn find(file: &str, env: &str) -> Option<String> { /// Clang libraries required to link to libclang statically. const CLANG_LIBRARIES: &'static [&'static str] = &[ "clang", - "clangARCMigrate", ...
1
0
8
d3d85ec5f3f4e45574a012ff2a0d282a49c24756
Kyle Mayes
2016-05-15T21:08:52
Refactor
diff --git a/build.rs b/build.rs index d21d4ca..051ab37 100644 --- a/build.rs +++ b/build.rs @@ -5,14 +5,19 @@ use std::process::{Command}; // Environment variables: // // * LLVM_CONFIG_PATH - provides a path to an `llvm-config` executable -// * LIBCLANG_PATH - provides a path to a libclang dynamic library -// * LIB...
1
31
21
bcc694fba2e1cf89eca9a4a96581178f39f34b82
Kyle Mayes
2016-05-11T01:21:34
Simplify build.rs
diff --git a/build.rs b/build.rs index 0234df8..6445d43 100644 --- a/build.rs +++ b/build.rs @@ -30,13 +30,15 @@ fn error(prefix: &str) -> ! { panic!("{:?}, set the LIBCLANG_PATH environment variable to your system's libclang directory", prefix); } +/// Runs a console command, returning the output if the comman...
1
11
13
c427881a0bcc8df491e20d591a8c6b417a18a848
Kyle Mayes
2016-05-10T18:32:11
Fix target_os value for OS X
diff --git a/build.rs b/build.rs index e10d335..faf9d92 100644 --- a/build.rs +++ b/build.rs @@ -44,7 +44,7 @@ fn find_libclang() -> Option<(String, String)> { } else { if cfg!(any(target_os="freebsd", target_os="linux")) { SEARCH_LINUX - } else if cfg!(target_os="osx") { + } el...
1
1
1
8eed98fc9da3996718bb68e553db3b383c4036c3
Kyle Mayes
2016-05-10T18:31:47
Reformat
diff --git a/build.rs b/build.rs index 8a10070..e10d335 100644 --- a/build.rs +++ b/build.rs @@ -40,10 +40,7 @@ fn find_libclang() -> Option<(String, String)> { let search = if let Ok(directory) = env::var("LIBCLANG_PATH") { vec![directory] } else if let Some(output) = run("llvm-config", &["--libdir"...
1
3
14
3ad0dc71e3696e2968faeb8fe8828855999d96cd
Yamakaky
2016-05-10T18:29:34
Improve find_libclang (#11)
diff --git a/build.rs b/build.rs index 2fc45e9..8a10070 100644 --- a/build.rs +++ b/build.rs @@ -39,32 +39,33 @@ fn run(command: &str, arguments: &[&str]) -> Option<String> { fn find_libclang() -> Option<(String, String)> { let search = if let Ok(directory) = env::var("LIBCLANG_PATH") { vec![directory] +...
1
23
22
6ca6f9f53313678bfeaa10424e2ea13e764819b2
Kyle Mayes
2016-05-09T23:17:27
Readd Linux backup search paths
diff --git a/build.rs b/build.rs index e618603..2fc45e9 100644 --- a/build.rs +++ b/build.rs @@ -2,6 +2,19 @@ use std::env; use std::path::{Path}; use std::process::{Command}; +const SEARCH_LINUX: &'static [&'static str] = &[ + "/usr/lib", + "/usr/lib/llvm", + "/usr/lib/llvm-3.8/lib", + "/usr/lib/llvm-3...
1
17
6
cb0b2e521201bfd67f54d24cbe1e749c6b1e5508
Kyle Mayes
2016-05-09T23:02:48
Improve build.rs error messages
diff --git a/build.rs b/build.rs index 1b77fd3..0e12544 100644 --- a/build.rs +++ b/build.rs @@ -13,6 +13,10 @@ const SEARCH_WINDOWS: &'static [&'static str] = &[ "C:\\Program Files\\LLVM\\lib", ]; +fn error(prefix: &str) -> ! { + panic!("{:?}, set the LIBCLANG_PATH environment variable to your system's libc...
1
15
9
33f6615a6cf52d984accda0d3c5d40e00ec54b2c
Yamakaky
2016-05-09T22:54:20
Remove rustc_version dependency (#8)
diff --git a/Cargo.toml b/Cargo.toml index 3724dac..5b17d51 100644 --- a/Cargo.toml +++ b/Cargo.toml @@ -32,7 +32,3 @@ static = [] bitflags = "0.6.0" libc = "0.2.11" - -[build-dependencies] - -rustc_version = "0.1.7" diff --git a/build.rs b/build.rs index 455a4bf..1b77fd3 100644 --- a/build.rs +++ b/build.rs @@ -1,...
2
0
6
1c513da3dc7ad082116d630c5b5b306ff1640c0a
Yamakaky
2016-05-09T22:48:45
Improve build (#7) Improve build.rs
diff --git a/build.rs b/build.rs index 5607584..455a4bf 100644 --- a/build.rs +++ b/build.rs @@ -4,19 +4,6 @@ use std::env; use std::path::{Path}; use std::process::{Command}; -const SEARCH_LINUX: &'static [&'static str] = &[ - "/usr/lib", - "/usr/lib/llvm", - "/usr/lib/llvm-3.8/lib", - "/usr/lib/llvm-3...
1
26
59
28af5f3a2a4cc7aaaa28155812f37a1093eeea82
Kyle Mayes
2016-03-16T20:21:09
Bump libc version
diff --git a/Cargo.toml b/Cargo.toml index 25311fb..b7bf101 100644 --- a/Cargo.toml +++ b/Cargo.toml @@ -30,7 +30,7 @@ gte_clang_3_8 = [] [dependencies] bitflags = "0.5.0" -libc = "0.2.7" +libc = "0.2.8" [build-dependencies]
1
1
1
c0435327c4bcb8437bd3dd53259943e0559974ed
Kyle Mayes
2016-03-16T20:20:20
Bump bitflags version
diff --git a/Cargo.toml b/Cargo.toml index 5b487bc..25311fb 100644 --- a/Cargo.toml +++ b/Cargo.toml @@ -29,7 +29,7 @@ gte_clang_3_8 = [] [dependencies] -bitflags = "0.4.0" +bitflags = "0.5.0" libc = "0.2.7" [build-dependencies] diff --git a/src/lib.rs b/src/lib.rs index 3d9dcf7..09298b4 100644 --- a/src/lib.r...
2
14
14
cf4e8e62945a1cc6670b8e798cb1fb4f649af4db
Kyle Mayes
2016-02-14T00:21:13
Bump rustc_version version
diff --git a/Cargo.toml b/Cargo.toml index ce886c8..1adedda 100644 --- a/Cargo.toml +++ b/Cargo.toml @@ -34,4 +34,4 @@ libc = "0.2.7" [build-dependencies] -rustc_version = "0.1.4" +rustc_version = "0.1.6"
1
1
1
6282fcd50a38fa552d7989a89ef37998ef1e933e
Kyle Mayes
2016-02-14T00:21:04
Bump libc version
diff --git a/Cargo.toml b/Cargo.toml index 82f10ac..ce886c8 100644 --- a/Cargo.toml +++ b/Cargo.toml @@ -30,7 +30,7 @@ gte_clang_3_8 = [] [dependencies] bitflags = "0.4.0" -libc = "0.2.4" +libc = "0.2.7" [build-dependencies]
1
1
1
223b01e3caf4eed4fe3f758633c268eba1ffb415
Kyle Mayes
2016-02-14T00:20:54
Bump bitflags version
diff --git a/Cargo.toml b/Cargo.toml index fe13fce..82f10ac 100644 --- a/Cargo.toml +++ b/Cargo.toml @@ -29,7 +29,7 @@ gte_clang_3_8 = [] [dependencies] -bitflags = "0.3.3" +bitflags = "0.4.0" libc = "0.2.4" [build-dependencies]
1
1
1
5178d195b750b2dd471db9ba53e8a36f84801d72
Kyle Mayes
2015-12-30T02:11:06
Remove unnecessary lint attributes
diff --git a/src/lib.rs b/src/lib.rs index ff98b43..4cf1ec3 100644 --- a/src/lib.rs +++ b/src/lib.rs @@ -9,7 +9,7 @@ //! The documentation for the non-version specific API is //! [here](https://kylemayes.github.io/clang-sys/all/clang_sys). -#![allow(missing_debug_implementations, missing_docs, non_upper_case_global...
1
1
1
5b53a23bcfdadd4ab17b53db3d7851ddddf33f84
Kyle Mayes
2015-12-26T17:36:40
Bump version
diff --git a/Cargo.toml b/Cargo.toml index 2e0a363..40b65d2 100644 --- a/Cargo.toml +++ b/Cargo.toml @@ -3,7 +3,7 @@ name = "clang-sys" authors = ["Kyle Mayes <kyle@mayeses.com>"] -version = "0.1.0" +version = "0.1.1" readme = "README.md" license = "MIT"
1
1
1
ff234140412e9e57326ad65f06addec8d8d78266
Andrew Gallant
2026-03-03T23:28:02
jiff-tzdb-0.1.6
diff --git a/crates/jiff-tzdb/Cargo.toml b/crates/jiff-tzdb/Cargo.toml index 7330144..8d87683 100644 --- a/crates/jiff-tzdb/Cargo.toml +++ b/crates/jiff-tzdb/Cargo.toml @@ -1,6 +1,6 @@ [package] name = "jiff-tzdb" -version = "0.1.5" #:version +version = "0.1.6" #:version authors = ["Andrew Gallant <jamslam@gmail.c...
1
1
1
c3d960e1922b9c88bfd8e50f061df4987503c4dd
Andrew Gallant
2026-02-22T20:32:50
fuzz: update dependencies
diff --git a/fuzz/Cargo.lock b/fuzz/Cargo.lock index 1f8d01d..f3f1d14 100644 --- a/fuzz/Cargo.lock +++ b/fuzz/Cargo.lock @@ -13,9 +13,9 @@ dependencies = [ [[package]] name = "cc" -version = "1.2.55" +version = "1.2.56" source = "registry+https://github.com/rust-lang/crates.io-index" -checksum = "47b26a0954ae34af0...
1
10
10
9e766682f9347a26ba9d3a97ea4d75427a60c645
Dennis Waldherr
2026-02-22T20:29:34
civil: check for `days < 1` in `Date::new` Fixes #523
diff --git a/src/civil/date.rs b/src/civil/date.rs index fc90f0e..739f934 100644 --- a/src/civil/date.rs +++ b/src/civil/date.rs @@ -246,7 +246,7 @@ impl Date { pub fn new(year: i16, month: i8, day: i8) -> Result<Date, Error> { let year = b::Year::check(year)?; let month = b::Month::check(month)?...
1
17
1
d57657ab903fa94eb17ce669605ee3116a0070ec
Ben Joffe
2026-02-18T01:19:09
perf: optimize timestamp -> datetime conversion Essentially, division and remainder operations are a lot slower when dealing with signed integers than unsigned. By using a positive offset bias in the ITimestamp to_datetime function, to convert the seconds to unsigned, I have found that the timestamp/to_civil_datetime_...
diff --git a/crates/jiff-static/src/shared/util/itime.rs b/crates/jiff-static/src/shared/util/itime.rs index b320234..2009524 100644 --- a/crates/jiff-static/src/shared/util/itime.rs +++ b/crates/jiff-static/src/shared/util/itime.rs @@ -49,9 +49,19 @@ impl ITimestamp { #[cfg_attr(feature = "perf-inline", inline(al...
2
156
6
b6977a12f2bba57b128e351bc307de8c924f9cfa
Andrew Gallant
2026-02-11T16:59:10
examples: add program demonstrating Diesel footgun with `TIMESTAMP WITHOUT TIME ZONE` values I wrote this example to demonstrate precisely why it's inappropriate for Jiff to provide trait implementations permitting interoperation between `jiff::Timestamp` and `TIMESTAMP WITHOUT TIME ZONE` values in PostgreSQL. Ref #5...
diff --git a/examples/diesel-postgres-timestamp-confusion/Cargo.toml b/examples/diesel-postgres-timestamp-confusion/Cargo.toml new file mode 100644 index 0000000..b0b2c96 --- /dev/null +++ b/examples/diesel-postgres-timestamp-confusion/Cargo.toml @@ -0,0 +1,20 @@ +[package] +publish = false +name = "diesel-postgres-tim...
3
168
2
4c33013130c19211e46db561e233666bacf69309
Andrew Gallant
2026-02-09T17:32:43
cargo: remove `js` feature from Playground metadata OK, seems like we might not want or need it: https://github.com/BurntSushi/jiff/issues/502#issuecomment-3872930901
diff --git a/Cargo.toml b/Cargo.toml index fb5bd8d..8ab6888 100644 --- a/Cargo.toml +++ b/Cargo.toml @@ -291,7 +291,7 @@ rustdoc-args = ["--cfg", "docsrs_jiff"] # Ref: https://github.com/BurntSushi/jiff/issues/502 [package.metadata.playground] default-features = true -features = ["js", "logging", "serde", "static", ...
1
1
1
ea994edcf44054f5b6a392131b8fa3321b8f6ec5
Andrew Gallant
2026-02-02T20:00:49
zoned: rip ranged integers out of time zone aware datetimes
diff --git a/src/util/b.rs b/src/util/b.rs index 17163f5..a438d73 100644 --- a/src/util/b.rs +++ b/src/util/b.rs @@ -360,6 +360,26 @@ define_bounds! { (YearCE, i16, "CE year", 1, Year::MAX), (YearBCE, i16, "BCE year", 1, Year::MAX + 1), (YearTwoDigit, i16, "year (2 digits)", 0, 99), + ( + Zoned...
5
68
122
8917c86eb2ea55550c7d08ed173089778b9efc78
Andrew Gallant
2026-02-02T00:46:59
civil: rip ranged integers out of `civil::DateTime`
diff --git a/src/civil/datetime.rs b/src/civil/datetime.rs index b57874b..e9b1180 100644 --- a/src/civil/datetime.rs +++ b/src/civil/datetime.rs @@ -12,11 +12,7 @@ use crate::{ }, shared::util::itime::IDateTime, tz::TimeZone, - util::{ - rangeint::{RFrom, RInto}, - round::increment, - ...
3
62
94
ea566c87cd33c42b0c1610dcc47345e52ab96bd3
Andrew Gallant
2026-01-23T19:59:32
civil: rip ranged integers out of `civil::Weekday`
diff --git a/src/civil/date.rs b/src/civil/date.rs index ebbcbe7..45a16ce 100644 --- a/src/civil/date.rs +++ b/src/civil/date.rs @@ -344,9 +344,9 @@ impl Date { UnixEpochDay::borked(iso_week_start_from_year(weekdate.year())); let year = t::NoUnits16::borked(weekdate.year()); let week = t:...
5
69
130
b151db3ff9becc8707e904f83875e2756ee80046
Andrew Gallant
2026-01-20T01:33:55
civil: rip ranged integers out of `civil::ISOWeekDate` The code is in a bit of a mess. I'll loop back around to do some clean-up after ranged integers are completely gone.
diff --git a/src/civil/date.rs b/src/civil/date.rs index db5b80f..ebbcbe7 100644 --- a/src/civil/date.rs +++ b/src/civil/date.rs @@ -11,6 +11,7 @@ use crate::{ shared::util::itime::{self, IDate, IEpochDay}, tz::TimeZone, util::{ + b, rangeint::{self, Composite, RFrom, RInto, TryRFrom}, ...
5
138
234
6bd2c0997f18f21cbb1fab812ae39da70a0a9351
Andrew Gallant
2026-01-19T19:19:50
rangeint: various other small removals of ranged integers I think these are the last remaining low hanging fruits.
diff --git a/src/lib.rs b/src/lib.rs index f285640..06336ea 100644 --- a/src/lib.rs +++ b/src/lib.rs @@ -812,38 +812,38 @@ mod tests { #[cfg(feature = "std")] #[test] fn ranges() { - use crate::util::t; - - dbg!((t::SpanYears::MIN, t::SpanYears::MAX)); - dbg!((t::SpanMonths::MIN, t::...
4
47
32
9c6306ef7404298dd287f2b2791dd369d2e334cb
Andrew Gallant
2026-01-19T18:50:38
span: remove `b::Sign::as_ranged_integer()` We just inline the implementation since it's only used once. We'll get get rid of `t::Sign` eventually.
diff --git a/src/fmt/util.rs b/src/fmt/util.rs index 95d32d7..391e64d 100644 --- a/src/fmt/util.rs +++ b/src/fmt/util.rs @@ -233,7 +233,7 @@ impl DurationUnits { // The unchecked setters above don't manipulate // the sign, which defaults to zero. So we need to // set it even when it's positiv...
3
8
11
e9412fb111e0e2b2e96e03e3ebdd7bd54a9ba20a
Andrew Gallant
2026-01-17T15:46:16
round: implement the core rounding primitives without ranged integers I tried to switch them more toward using `SignedDuration`, but because increment is an `i64`, it's possible to overflow `SignedDuration`. So we stick to `i128` for the time being. (I kind of wonder if we should switch `increment` to `i32`, but that...
diff --git a/src/civil/datetime.rs b/src/civil/datetime.rs index 5bdb347..28e11ea 100644 --- a/src/civil/datetime.rs +++ b/src/civil/datetime.rs @@ -3523,7 +3523,7 @@ impl DateTimeRound { // ref: https://tc39.es/proposal-temporal/#sec-temporal.plaindatetime.prototype.round let increment = - ...
10
251
263
1eecb4714462f0e75f1606039abce803a54df623
Andrew Gallant
2026-01-17T02:20:13
span: rewrite `Span::from_invariant_seconds` We now use a `SignedDuration`, which means we can farm out the math to it. (I expect this to be a recurring refactoring pattern in the near term.) This also simplifies the logic and drops LLVM lines by ~700 in my Biff benchark. I wouldn't be surprised if this were faster t...
diff --git a/src/signed_duration.rs b/src/signed_duration.rs index 8ad02d9..a97d3f5 100644 --- a/src/signed_duration.rs +++ b/src/signed_duration.rs @@ -2327,6 +2327,54 @@ impl SignedDuration { SignedDuration::from_secs((mins as i64) * SECS_PER_MINUTE) } + /// Returns the number of whole civil weeks ...
4
100
261
f598b82db796dc3cc21f6bf167dda769a7021916
Andrew Gallant
2026-01-16T23:10:57
span: simplify `Span::from_invariant_nanoseconds` ... with the idea of switching it to use `SignedDuration`. This might also reduce code size? Indeed, in my Biff benchmark, `Span::from_invariant_nanoseconds` reduces from 767 LLVM lines to 320 lines.
diff --git a/src/span.rs b/src/span.rs index 3d6695e..b2bf25a 100644 --- a/src/span.rs +++ b/src/span.rs @@ -2878,148 +2878,54 @@ impl Span { nanos: NoUnits128, ) -> Result<Span, Error> { let mut span = Span::new(); - match largest { - Unit::Week => { - let micros...
1
47
141
73f1721812f355066976bd4f8ee1d39570a30a69
Andrew Gallant
2026-01-16T16:12:06
timestamp: optimize adding `Span` to `Timestamp` The `timestamp/add_time_secs/span/jiff` benchmark regressed a little bit from the previous commit. It looks like something changed that caused more copying of `Span` values. In this commit, we regain that perf by being more explicit with `&Span`.
diff --git a/bench/src/timestamp.rs b/bench/src/timestamp.rs index 33fb01f..75a66d0 100644 --- a/bench/src/timestamp.rs +++ b/bench/src/timestamp.rs @@ -12,6 +12,7 @@ use crate::{benchmark, convert::ConvertFrom}; pub(super) fn define(c: &mut Criterion) { add_time_secs(c); + sub_time_secs(c); add_time_su...
10
91
56
c1e922e89933bd2fc7c0d61163e729cdc208dbda
Andrew Gallant
2026-01-14T21:45:35
tz: rip ranged integers out of `jiff::tz::Offset` implementation This was a little harder, but still not too bad. This also begins starting to rip ranged integers out of some other places, including beginning to reduce the uses of the cursed `Composite` API used for converting to and from values which use multiple ran...
diff --git a/src/civil/datetime.rs b/src/civil/datetime.rs index f44d1ec..a6a932e 100644 --- a/src/civil/datetime.rs +++ b/src/civil/datetime.rs @@ -13,7 +13,7 @@ use crate::{ shared::util::itime::IDateTime, tz::TimeZone, util::{ - rangeint::{Composite, RFrom, RInto}, + rangeint::{RFrom, RI...
12
464
226
d89d40a4e23381adf57e1d6bd8bb5740021c46ce
Andrew Gallant
2026-01-13T02:15:05
util: use named constants more consistently At least, for defining the boundaries. There are more places where we don't use "basic" constants and still write out "magic" numbers. Usually they are very well known though, so I don't feel too bad about it.
diff --git a/src/util/b.rs b/src/util/b.rs index 2276184..15a2c33 100644 --- a/src/util/b.rs +++ b/src/util/b.rs @@ -9,22 +9,32 @@ ranged integers. I'm not quite sure where this will go. use crate::{util::t, Error}; +pub(crate) const DAYS_PER_WEEK: i64 = 7; +pub(crate) const HOURS_PER_CIVIL_DAY: i64 = 24; pub(cra...
1
29
19
340baa547b4cf5efa787cd3eac18954e36df1774
Andrew Gallant
2026-01-13T02:06:25
tz: rip ranged integers out of timezone representation internals This was easy.
diff --git a/src/tz/timezone.rs b/src/tz/timezone.rs index 9c2e4e8..f06f8d1 100644 --- a/src/tz/timezone.rs +++ b/src/tz/timezone.rs @@ -1938,10 +1938,7 @@ impl<'t> TimeZoneAbbreviation<'t> { mod repr { use core::mem::ManuallyDrop; - use crate::{ - tz::tzif::TzifStatic, - util::{constant::unwra...
1
3
7
23c879a5881a7c605627818f6f7ee1635eeb95f3
Andrew Gallant
2026-01-13T02:01:48
tz: rip ranged integers out of Zic parser It's so sad keeping this up to date. I haven't touched it since my initial push to build a Zic compiler. I'd love to revisit it and complete it... some day.
diff --git a/src/tz/zic.rs b/src/tz/zic.rs index dc702f3..73a240d 100644 --- a/src/tz/zic.rs +++ b/src/tz/zic.rs @@ -98,7 +98,7 @@ in such cases. [Time Zone Database]: https://www.iana.org/time-zones */ -#![allow(warnings)] +#![allow(dead_code)] use core::{ops::RangeInclusive, str::FromStr}; @@ -118,12 +118,7 ...
1
131
168
77ec9cf3fea7ec16bba8c56d8c972b34393d1803
Andrew Gallant
2026-02-07T18:59:36
cargo: add `js` feature to Rust Playground deployment Because it's being used already: https://github.com/rust-lang/rust-playground/blob/f0d85ff3b04992974c8ef4ff72be4f27c47ec7c5/compiler/base/Cargo.toml#L899-L903 This also means that enabling `tz-system` should work. That is, it should fetch the "system" time zone fr...
diff --git a/Cargo.toml b/Cargo.toml index 8ab6888..fb5bd8d 100644 --- a/Cargo.toml +++ b/Cargo.toml @@ -291,7 +291,7 @@ rustdoc-args = ["--cfg", "docsrs_jiff"] # Ref: https://github.com/BurntSushi/jiff/issues/502 [package.metadata.playground] default-features = true -features = ["logging", "serde", "static", "tzdb-...
1
1
1
3c9ba667390a1f3aeb104fb9110475defa609bb0
Andrew Gallant
2026-02-01T17:23:49
zoned: kinda optimize `Zoned::default()` This adds a benchmark that just calls `Zoned::default()` and internally changes its implementation to return a `const` value. Interestingly, the benchmark measurement result remains unchanged. It seems like the compiler was actually seeing through the `Zoned::new(Timestamp::de...
diff --git a/bench/src/zoned.rs b/bench/src/zoned.rs index 6d02af4..cb34192 100644 --- a/bench/src/zoned.rs +++ b/bench/src/zoned.rs @@ -10,11 +10,43 @@ use jiff::{ use crate::{benchmark, convert::ConvertFrom}; pub(super) fn define(c: &mut Criterion) { + default(c); fixed_offset_add_time(c); fixed_offs...
2
49
19
66ba0fe164c91082d63bb2687c41629bb125bd26
Andrew Gallant
2026-01-12T22:04:58
fmt: rip ranged integers out of Temporal RFC 9557 parser The `jiff::fmt` module is now completely free of ranged integer types. w00t. Easier than I expected, but I also suspected this would be one of the easier parts of the migration since `jiff::fmt` lives at the boundaries.
diff --git a/src/error/fmt/temporal.rs b/src/error/fmt/temporal.rs index 4f0cd34..d94fbfe 100644 --- a/src/error/fmt/temporal.rs +++ b/src/error/fmt/temporal.rs @@ -63,17 +63,9 @@ pub(crate) enum Error { FailedYearInDate, FailedYearInYearMonth, InvalidDate, - InvalidDay, - InvalidHour, - Invalid...
4
89
126
a651ac732666cd4ebe0a5dbc2b5dcde0088632cf
Andrew Gallant
2026-01-12T22:00:11
error: update `Error::is_range` predicate This should now include the new `Bounds` error type.
diff --git a/src/error/mod.rs b/src/error/mod.rs index 8406a5e..e72ae14 100644 --- a/src/error/mod.rs +++ b/src/error/mod.rs @@ -111,7 +111,10 @@ impl Error { /// ``` pub fn is_range(&self) -> bool { use self::ErrorKind::*; - matches!(*self.root().kind(), Range(_) | SlimRange(_) | ITimeRange(_...
1
4
1
c2dcbca9f8a2619e130581d5b787c6de9a3e55e2
Andrew Gallant
2026-01-12T21:05:50
fmt: refactor Temporal parser to get rid of input capture I did this originally to aid error messages by reflecting back precisely what was in the input. But I've been trending away from that, and we should be able to get most of the value here by just using the `Display` impls of what we parsed. That's what we do her...
diff --git a/src/fmt/offset.rs b/src/fmt/offset.rs index 9d6df16..149a627 100644 --- a/src/fmt/offset.rs +++ b/src/fmt/offset.rs @@ -181,6 +181,17 @@ impl ParsedOffset { } } +impl core::fmt::Display for ParsedOffset { + fn fmt(&self, f: &mut core::fmt::Formatter) -> core::fmt::Result { + match self.ki...
3
120
160
d439a66e87e3e62e5b6ca4d86a5353fd5a500b9c
Andrew Gallant
2026-01-11T22:12:52
fmt: rename `strtime::format` to `strtime::print` This was driving be bonkers. Everything else uses `print.rs` (when parsing and printing are in separate modules).
diff --git a/src/fmt/strtime/mod.rs b/src/fmt/strtime/mod.rs index acc085b..383be5b 100644 --- a/src/fmt/strtime/mod.rs +++ b/src/fmt/strtime/mod.rs @@ -279,7 +279,7 @@ use crate::{ error::{fmt::strtime::Error as E, ErrorContext}, fmt::{ buffer::{ArrayBuffer, BorrowedWriter}, - strtime::{forma...
2
2
2
16da5e8ed939e959333822172a5e2e6b87290c5d
Andrew Gallant
2026-01-11T22:05:53
fmt: rip ranged integers out of time zone offset parser This also rips them out of the utility functions in `fmt`. (This was already intentionally quite light on ranged integer usage, since I had written that code recently and figured I'd be ripping things out.) This did require defining more bounds types, including, ...
diff --git a/src/error/fmt/offset.rs b/src/error/fmt/offset.rs index af81804..785633b 100644 --- a/src/error/fmt/offset.rs +++ b/src/error/fmt/offset.rs @@ -21,9 +21,6 @@ pub(crate) enum Error { ParseMinutes, ParseSeconds, PrecisionLoss, - RangeHours, - RangeMinutes, - RangeSeconds, Separat...
5
240
119
05f991a79315e2315064475824556ca06ed1ed26
Andrew Gallant
2026-01-11T19:55:57
fmt: rip ranged integers out of RFC 2822 parsing This continues the journey started in 808905c5.
diff --git a/src/error/fmt/rfc2822.rs b/src/error/fmt/rfc2822.rs index 764b267..ab2ab16 100644 --- a/src/error/fmt/rfc2822.rs +++ b/src/error/fmt/rfc2822.rs @@ -19,15 +19,9 @@ pub(crate) enum Error { FailedZoned, InconsistentWeekday { parsed: Weekday, from_date: Weekday }, InvalidDate, - InvalidHour, ...
3
180
77
bd32085efeb12ba22871c939200ab25358dc5d64
Andrew Gallant
2026-01-10T01:34:43
fuzz: bump dependencies
diff --git a/fuzz/Cargo.lock b/fuzz/Cargo.lock index 8f2fadd..8edaaed 100644 --- a/fuzz/Cargo.lock +++ b/fuzz/Cargo.lock @@ -13,9 +13,9 @@ dependencies = [ [[package]] name = "cc" -version = "1.2.50" +version = "1.2.52" source = "registry+https://github.com/rust-lang/crates.io-index" -checksum = "9f50d563227a1c37c...
1
16
16
efe4616a2a0b956ea1db1b5beaa78aaed30339fd
Andrew Gallant
2026-01-06T22:09:44
rangeint: switch `/` and `%` operators to match std When I first devised the rangeint abstraction, I somewhat unwisely made its `/` and `%` operators use eucilidean division instead of matching what `std` does (which always rounds the result toward zero). I did this because I thought it was going to be overwhelmingly ...
diff --git a/src/civil/date.rs b/src/civil/date.rs index 2dc1a73..42535df 100644 --- a/src/civil/date.rs +++ b/src/civil/date.rs @@ -3654,8 +3654,8 @@ fn month_add_overflowing( ) -> (t::Month, t::SpanYears) { let month = t::SpanMonths::rfrom(month); let total = month - C(1) + span; - let years = total / C...
5
41
18
d8e765bd78c60f60e9320f021e3560e1a47741a4
Andrew Gallant
2026-01-04T21:07:51
fmt: use `impl Into<u64>` for some routines on `BorrowedBuffer` I had previously not done this thinking that it would be advantageous for these routines to be monomorphic. But in the end, these all end up getting forced to be inlined, which means we might as well make the call sites more ergonomic. This doesn't have ...
diff --git a/src/fmt/buffer.rs b/src/fmt/buffer.rs index cc10031..fb5176e 100644 --- a/src/fmt/buffer.rs +++ b/src/fmt/buffer.rs @@ -243,7 +243,8 @@ impl<'data> BorrowedBuffer<'data> { /// When the available space is insufficient to encode the number of /// digits in the decimal representation of `n`. #[...
6
104
116
79de86aa86faf157de15da0201f36d828eeb6add
Andrew Gallant
2026-01-04T20:51:04
fmt: improve SAFETY comments I switched up the implementation here a few commits ago, but forgot to update the SAFETY comments.
diff --git a/src/fmt/buffer.rs b/src/fmt/buffer.rs index f405d9c..cc10031 100644 --- a/src/fmt/buffer.rs +++ b/src/fmt/buffer.rs @@ -391,7 +391,10 @@ impl<'data> BorrowedBuffer<'data> { .get_mut(..2) .expect("padded 2 digit integer exceeds available buffer space"); let radix_offset = ...
1
16
4
52ad94e9490d51fad230c5d24ce1ee1c292e2ee0
Andrew Gallant
2026-01-04T20:49:06
fmt: use `MaybeUninit::write` in lieu of `MaybeUninit::new` I forgot about the `write` API, which is often more convenient.
diff --git a/src/fmt/buffer.rs b/src/fmt/buffer.rs index b58ecb0..f405d9c 100644 --- a/src/fmt/buffer.rs +++ b/src/fmt/buffer.rs @@ -229,11 +229,10 @@ impl<'data> BorrowedBuffer<'data> { #[cfg_attr(feature = "perf-inline", inline(always))] pub(crate) fn write_ascii_char(&mut self, byte: u8) { assert!...
1
14
14
1cc43ff1d6e874c1d12e695f028bfd14b359c8df
tison
2026-01-04T19:52:25
doc: fix typo in `SPAN_CIVIL_SECONDS` comment PR #473
diff --git a/src/util/t.rs b/src/util/t.rs index 96a589a..71f8c0c 100644 --- a/src/util/t.rs +++ b/src/util/t.rs @@ -278,7 +278,7 @@ pub(crate) type SpanMinutes = /// /// Note that neither the existence of this constant nor defining our spans /// based on it impacts the correctness of doing arithmetic on zoned insta...
1
1
1
6b728036c6f582105c64f44c92f67ec695896542
tison
2026-01-04T19:52:00
impl: avoid superfluous clone of `RelativeZoned` PR #472
diff --git a/src/span.rs b/src/span.rs index 967052a..7489889 100644 --- a/src/span.rs +++ b/src/span.rs @@ -6578,8 +6578,7 @@ impl Nudge { let mut rounded_time_nanos = mode.round_by_unit_in_nanoseconds(time_nanos, smallest, increment); let (relative0, relative1) = clamp_relative_span( - ...
1
1
2
891380d69d0f528d83dbdc7db1eff4b9c60b0d1c
Andrew Gallant
2026-01-04T19:26:14
bench: add benchmark for `time-format` It's extremely slow: ``` $ critcmp x01 -g '(.*/)(?:jiff|time-format)$' -f print/strftime/oneshot/to_string group x01/jiff x01/time-format ----- -------- ----...
diff --git a/bench/Cargo.toml b/bench/Cargo.toml index f18af00..7aac919 100644 --- a/bench/Cargo.toml +++ b/bench/Cargo.toml @@ -31,6 +31,8 @@ time = { version = "0.3.36", features = ["formatting", "macros", "parsing"] } humantime = "2.1.0" +time-format = "1.2.2" + [profile.release] debug = true diff --git a/b...
2
22
0
64dc8daee7765bb69f608f684f56716df04f5c75
Andrew Gallant
2026-01-04T17:54:14
fmt: optimize integer formatting This gets another small win by optimizing small integer formatting: ``` $ critcmp baseline pr pr2 pr3 -f jiff group baseline pr pr2 pr3 ----- ...
diff --git a/src/fmt/buffer.rs b/src/fmt/buffer.rs index a24b1e8..b58ecb0 100644 --- a/src/fmt/buffer.rs +++ b/src/fmt/buffer.rs @@ -13,6 +13,30 @@ const MAX_PRECISION: usize = 9; /// This relies on the fact that the maximum padding is clamped to `20`. const BROAD_MINIMUM_BUFFER_LEN: usize = 20; +/// All integers i...
2
85
21
e7cb6473a23cba0c9c4d1bd686af862f67362703
Andrew Gallant
2026-01-03T13:43:39
fmt: further refactor `strftime` This is another improvement on runtime performance *and* a decrease in binary size. For binary size, we now get a 27,000 LLVM line improvement on my synthetic benchmark (up from a 23,000 line improvement in the previous commit) and a 14,000 LLVM line improvement on my Biff benchmark (u...
diff --git a/src/fmt/strtime/format.rs b/src/fmt/strtime/format.rs index 544ae33..5c11f3c 100644 --- a/src/fmt/strtime/format.rs +++ b/src/fmt/strtime/format.rs @@ -8,7 +8,7 @@ use crate::{ strtime::{ month_name_abbrev, month_name_full, weekday_name_abbrev, weekday_name_full, BrokenDo...
3
438
337
317f81e96cd9b3dcf50063567818243434ee1d87
Andrew Gallant
2026-01-03T01:57:48
fmt: refactor `strftime` to use uninitialized buffers The code size decrease in my synthetic benchmark is quite substantial: it reduces LLVM line count by around 23,000 lines. But there's effectively no change in Biff LLVM lines. Weird. I don't understand it. Biff does use two different implementations of `Custom`, so...
diff --git a/src/fmt/buffer.rs b/src/fmt/buffer.rs index a805523..a24b1e8 100644 --- a/src/fmt/buffer.rs +++ b/src/fmt/buffer.rs @@ -186,6 +186,22 @@ impl<'data> BorrowedBuffer<'data> { self.filled += string.len() as u16; } + /// Writes the given Unicode scalar value (as UTF-8) to this writer. + /...
6
114
476
fe9b914277b56c1f1d73ca81bb29794bfba0ac8a
Andrew Gallant
2026-01-02T23:16:27
bench: add more strftime benchmarks We don't really vary up the format string (arguably we should), but instead add more coverage for other APIs. Current results: ``` $ critcmp baseline -g '(.*/)(?:jiff|time|chrono)$' group baseline/chrono baseline/jiff ...
diff --git a/bench/src/print.rs b/bench/src/print.rs index 7d27967..75bc662 100644 --- a/bench/src/print.rs +++ b/bench/src/print.rs @@ -823,6 +823,9 @@ fn print_strftime(c: &mut Criterion) { // `strftime`-style APIs, but I decided to just copy this from the existing // `strptime` benchmark for now. (Which wa...
1
81
4
e683626b0e6c8db94379c7edc02652bd1fd058aa
Andrew Gallant
2026-01-02T21:07:16
doc: add panic conditions for `Date::at` and `Time::on` This is just documented existing (and intended) behavior.
diff --git a/src/civil/date.rs b/src/civil/date.rs index e56f66f..2dc1a73 100644 --- a/src/civil/date.rs +++ b/src/civil/date.rs @@ -1307,6 +1307,19 @@ impl Date { /// A convenience function for constructing a [`DateTime`] from this date /// at the time given by its components. /// + /// # Panics + ...
2
28
0
e28b3a9a5205bda3326498ed5ca4005577a2d8a5
Andrew Gallant
2025-12-31T17:29:25
fmt: get rid of the internal `rfc9557` configuration knob It was always enabled. I believe I added this when I first wrote the printer, assuming it would need to be configurable. But it is, at time of writing, always enabled. So let's just squash it.
diff --git a/src/fmt/temporal/printer.rs b/src/fmt/temporal/printer.rs index cbf5f40..dd067c5 100644 --- a/src/fmt/temporal/printer.rs +++ b/src/fmt/temporal/printer.rs @@ -34,18 +34,12 @@ const MAX_DURATION_LEN: usize = 35; pub(super) struct DateTimePrinter { lowercase: bool, separator: u8, - rfc9557: bo...
1
5
18
7e32b136f6ffd7d81ab98f7bd8ff5b2484ca081e
Andrew Gallant
2025-12-31T16:50:32
fmt: tweak `BorrowedBuffer::with_writer` API Previously, we assumed that the size of the stack allocated buffer and the runtime allocated extra capacity would always be the same. This is okay for the case where there is very little difference between the minimum and maximum lengths. But for RFC 9557 zoned datetimes, t...
diff --git a/src/fmt/buffer.rs b/src/fmt/buffer.rs index 58ac3fe..4ec7f45 100644 --- a/src/fmt/buffer.rs +++ b/src/fmt/buffer.rs @@ -84,6 +84,7 @@ impl<'data> BorrowedBuffer<'data> { #[cfg_attr(feature = "perf-inline", inline(always))] pub(crate) fn with_writer<const N: usize>( wtr: &mut dyn Write, +...
2
5
1
97b5c17a1d56324216a6f47c960b95bd017bce48
Andrew Gallant
2025-12-31T20:29:56
bench: add RFC 3339 benchmarks for `humantime`
diff --git a/bench/src/print.rs b/bench/src/print.rs index 744642d..7d27967 100644 --- a/bench/src/print.rs +++ b/bench/src/print.rs @@ -1,4 +1,4 @@ -use std::hint::black_box as bb; +use std::{hint::black_box as bb, time::SystemTime}; use criterion::Criterion; use jiff::{ @@ -644,6 +644,7 @@ fn print_rfc3339(c: &mu...
1
26
1
c5ab0780698e205213d142f90f236f49f0a74eca
Andrew Gallant
2025-12-31T20:23:15
bench: move RFC 3339 printing benchmarks Just so that they are in lexicographic order. This doesn't make any other changes.
diff --git a/bench/src/print.rs b/bench/src/print.rs index c34c8e0..744642d 100644 --- a/bench/src/print.rs +++ b/bench/src/print.rs @@ -549,12 +549,12 @@ fn print_iso8601_duration(c: &mut Criterion) { } } -/// Measures the time it takes to print a `Zoned` in the RFC 3339 format. -fn print_rfc3339(c: &mut Crite...
1
41
41
91f5620e4e83fb779009d2977ab2705df0f3baac
Andrew Gallant
2025-12-31T00:51:11
fmt: refactor Temporal ISO 8601 duration printer to use uninitialized buffers In a simple benchmark program that prints spans, signed and unsigned durations using `jiff::fmt::temporal::SpanPrinter`, the total number of LLVM lines is reduced by about 6,000. There is also a modest improvement in runtime performance: `...
diff --git a/src/fmt/temporal/printer.rs b/src/fmt/temporal/printer.rs index 484bbfe..cbf5f40 100644 --- a/src/fmt/temporal/printer.rs +++ b/src/fmt/temporal/printer.rs @@ -1,20 +1,35 @@ +use core::time::Duration; + use crate::{ civil::{Date, DateTime, ISOWeekDate, Time}, error::{fmt::temporal::Error as E, E...
3
207
212
5f171b5c2005602f5c96f4f00218d791ccf3e589
Andrew Gallant
2025-12-30T23:47:35
span: replace some code with `SignedDuration` constructor This also reduces code size slightly.
diff --git a/src/span.rs b/src/span.rs index 57886ed..aa05d7f 100644 --- a/src/span.rs +++ b/src/span.rs @@ -2453,19 +2453,9 @@ impl Span { self.largest_unit() <= Unit::Week, "units must be weeks or lower" ); - - let seconds = nanos / t::NANOS_PER_SECOND; - let seconds =...
1
3
13
9b305467926bdf87a4918353a016d2b8270f3afc
tison
2025-12-30T15:32:04
sqlx: use `Timestamp::constant` for `POSTGRES_EPOCH_TIMESTAMP` I think this code was written before `Timestamp::constant` was added. PR #466
diff --git a/crates/jiff-sqlx/src/postgres.rs b/crates/jiff-sqlx/src/postgres.rs index 9d00a61..9557b7f 100644 --- a/crates/jiff-sqlx/src/postgres.rs +++ b/crates/jiff-sqlx/src/postgres.rs @@ -22,7 +22,8 @@ use crate::{Date, DateTime, Span, Time, Timestamp, ToSqlx}; static POSTGRES_EPOCH_DATE: civil::Date = civil::dat...
1
3
6
1a71c3041919ae7e2e7030c73c32f4ea38c0b590
Andrew Gallant
2025-12-29T18:13:26
fmt: refactor friendly printer to use padding/precision more explicitly Basically, instead of creating and passing around an `IntegerFormatter` or `FractionalFormatter`, we pass around padding and precision options explicitly. We then materialize the formatters as needed. While this looks odd, this is meant to prepare...
diff --git a/src/fmt/friendly/printer.rs b/src/fmt/friendly/printer.rs index 1a70f14..a4f5084 100644 --- a/src/fmt/friendly/printer.rs +++ b/src/fmt/friendly/printer.rs @@ -1316,18 +1316,22 @@ impl SpanPrinter { } } - let fmtint = - IntegerFormatter::new().padding(self.padding....
1
78
44
83e2f24213413035b2c552a2a972e2a0ce83f55a
Andrew Gallant
2025-12-29T17:55:48
fmt: small change to unexport `Fractional::new` I just find this makes it easier to see all of the entry points into the existing fractional formatter.
diff --git a/src/fmt/mod.rs b/src/fmt/mod.rs index 3097f65..5721462 100644 --- a/src/fmt/mod.rs +++ b/src/fmt/mod.rs @@ -507,7 +507,7 @@ trait WriteExt: Write { formatter: &FractionalFormatter, n: impl Into<u32>, ) -> Result<(), Error> { - self.write_fractional(&Fractional::new(formatter, ...
2
2
2
ff875c35120748f2101fc5d2bf134da2fe7bbf16
Andrew Gallant
2025-12-29T16:39:26
fmt: add more formatting routines to `BorrowedBuffer` We'll need these to port more of Jiff's printers to the uninitialized buffer abstraction. I'm somewhat unhappy with the extra code here, but all of these routines are intended to be inlined anyway. So this might end up saving code in the long run. I'm also somewh...
diff --git a/src/fmt/buffer.rs b/src/fmt/buffer.rs index 1b2e220..ad5bde5 100644 --- a/src/fmt/buffer.rs +++ b/src/fmt/buffer.rs @@ -3,10 +3,14 @@ use core::mem::MaybeUninit; use crate::{fmt::Write, Error}; const MAX_CAPACITY: usize = u8::MAX as usize; +// From `u64::MAX.to_string().len()`. +const MAX_INTEGER_LEN: ...
2
572
2
811802ca462651a49982f0c21a26124c770e773a
Andrew Gallant
2025-12-28T20:55:26
fmt: small tweak to friendly printer This reduces LLVM lines in my benchmark by a couple hundred lines.
diff --git a/src/fmt/friendly/printer.rs b/src/fmt/friendly/printer.rs index db866a3..1a70f14 100644 --- a/src/fmt/friendly/printer.rs +++ b/src/fmt/friendly/printer.rs @@ -1295,15 +1295,17 @@ impl SpanPrinter { span: &Span, mut wtr: W, ) -> Result<(), Error> { - let span_cal = span.only_c...
1
10
9
db9ba05f6c05905b0c20d4bd72e8f634bda842b1
Andrew Gallant
2025-12-28T13:40:41
fmt: clarify maximum padding/precision values There's no behavior changes here, but the maximums weren't previously documented.
diff --git a/src/fmt/friendly/printer.rs b/src/fmt/friendly/printer.rs index c2a9a07..db866a3 100644 --- a/src/fmt/friendly/printer.rs +++ b/src/fmt/friendly/printer.rs @@ -821,6 +821,9 @@ impl SpanPrinter { /// of `2` is used for units of hours, minutes and seconds. Otherwise, a /// padding of `0` is used. ...
2
7
1
57eed5e1f130381c492a4961c0c44220ff4654e1
Andrew Gallant
2025-12-28T01:28:42
bench: add more print benchmarks
diff --git a/bench/src/convert.rs b/bench/src/convert.rs index 7755ed5..26d69c8 100644 --- a/bench/src/convert.rs +++ b/bench/src/convert.rs @@ -95,6 +95,13 @@ impl ConvertFrom<jiff::civil::Time> for chrono::NaiveTime { } } +impl ConvertFrom<jiff::SignedDuration> for chrono::TimeDelta { + fn convert_from(x: ...
2
644
46
8849db3790c093f1328978d283763e213e072a7f
Andrew Gallant
2025-12-28T00:44:08
fmt: forcefully elide more bounds checks We're already using `unsafe` here, so we might as well go whole hog and eliminate as much code as possible.
diff --git a/src/fmt/buffer.rs b/src/fmt/buffer.rs index 0daae12..1b2e220 100644 --- a/src/fmt/buffer.rs +++ b/src/fmt/buffer.rs @@ -217,15 +217,19 @@ impl<'data> BorrowedBuffer<'data> { // a pain. let digits = if n == 0 { 1 } else { n.ilog10() as u8 + 1 }; let available = self.available(); +...
1
36
10
e41ded34f8d35af387258a56471c1d3da0189619
Andrew Gallant
2025-12-29T13:26:46
error: replace a few more instances of unnecessary `write!` calls I missed these because I had filtered `src/error/` out of my search for `write!` calls when looking for ones that could be removed. (Because `src/error/` is where `write!` is most commonly needed.)
diff --git a/src/error/mod.rs b/src/error/mod.rs index 47094da..9a21171 100644 --- a/src/error/mod.rs +++ b/src/error/mod.rs @@ -725,11 +725,11 @@ impl core::fmt::Display for IOError { fn fmt(&self, f: &mut core::fmt::Formatter) -> core::fmt::Result { #[cfg(feature = "std")] { - write!...
1
6
6
8f6a73fd32c8cd6b6b30414ae1c6a7c791204d40
Andrew Gallant
2025-12-28T13:17:35
signed_duration: update docs This is a bit of a follow-up to #462.
diff --git a/src/signed_duration.rs b/src/signed_duration.rs index 04edc9b..ad43610 100644 --- a/src/signed_duration.rs +++ b/src/signed_duration.rs @@ -150,13 +150,17 @@ const MINS_PER_HOUR: i64 = 60; /// duration, like [`SignedDuration::abs`] and [`SignedDuration::checked_neg`]. /// * A [`SignedDuration::system_unt...
1
6
2
25c9b335f70cd424f46d0eb357fad30d76f16943
Andrew Gallant
2025-12-26T01:54:25
error: squash warning on older versions of Rust Kind of interesting that this doesn't appear on newer versions of Rust. I think it's technically correct?
diff --git a/src/error/mod.rs b/src/error/mod.rs index c320a6d..47094da 100644 --- a/src/error/mod.rs +++ b/src/error/mod.rs @@ -696,8 +696,10 @@ impl core::fmt::Display for CrateFeatureError { TzdbZoneInfo => "tzdb-zoneinfo", }; #[allow(unreachable_code)] - core::fmt::Display::fmt...
1
4
2
2ed0e460a122912108eab42cb2a1e532ae703e4c
Andrew Gallant
2025-12-25T23:59:09
fmt: add support for a `&mut Vec<u8>` fast path The printers can use this method to write directly into the `Vec<u8>`'s spare capacity, instead of needing to write to our new uninitialized buffer and then copy the data via the `jiff::fmt::Write` interface. Using the uninitialized buffer without this is still a dramat...
diff --git a/src/fmt/mod.rs b/src/fmt/mod.rs index 1226ddf..3097f65 100644 --- a/src/fmt/mod.rs +++ b/src/fmt/mod.rs @@ -277,8 +277,9 @@ impl<'i, V: core::fmt::Debug> core::fmt::Debug for Parsed<'i, V> { /// The `std::fmt::Write` trait wasn't used itself because: /// /// 1. Using a custom trait allows us to require ...
1
49
2
dca73c5a5d20318ea186969ca357e1d051d61dcc
Andrew Gallant
2025-12-23T02:47:11
bench: add RFC 2822 printing benchmark We're going to be doing some surgery on the RFC 2822 printer. Mostly motivated by decreasing binary size. But we should be able to optimize runtime as well. The printer is actually already pretty slow compared to `time`: ``` $ critcmp base -g '(.*)/(?:humantime|jiff|chrono|time...
diff --git a/bench/src/print.rs b/bench/src/print.rs index db757c7..121b2cf 100644 --- a/bench/src/print.rs +++ b/bench/src/print.rs @@ -1,13 +1,18 @@ use std::hint::black_box as bb; use criterion::Criterion; -use jiff::civil; +use jiff::{ + civil, + tz::{Offset, TimeZone}, + Timestamp, +}; use crate::{...
1
96
1
65825cf1a34f603bbbdf4db20f9a513d0376f45e
Andrew Gallant
2025-12-24T19:32:23
fuzz: update dependencies
diff --git a/fuzz/Cargo.lock b/fuzz/Cargo.lock index 5e5046c..8f2fadd 100644 --- a/fuzz/Cargo.lock +++ b/fuzz/Cargo.lock @@ -13,9 +13,9 @@ dependencies = [ [[package]] name = "cc" -version = "1.2.49" +version = "1.2.50" source = "registry+https://github.com/rust-lang/crates.io-index" -checksum = "90583009037521a11...
1
6
6
edb7cd21422ae6b9131c447eb8685d77c55968ff
Andrew Gallant
2025-12-24T17:09:08
deps: bump jiff-tzdb to 0.1.5 This includes an updated to the tzdb 2025c release.
diff --git a/Cargo.toml b/Cargo.toml index 57a0974..ae48911 100644 --- a/Cargo.toml +++ b/Cargo.toml @@ -179,7 +179,7 @@ perf-inline = [] [dependencies] jiff-static = { version = "0.2", path = "crates/jiff-static", optional = true } -jiff-tzdb = { version = "0.1.4", path = "crates/jiff-tzdb", optional = true } +jif...
1
1
1
61dc9bd8aa1cec759a35983e2f5984f47e5d8bf7
Andrew Gallant
2025-12-23T01:46:31
fmt: rename `Decimal` to `Integer` And similarly for `DecimalFormatter`. I believe I originally called this `Decimal` because it was also used to do fractional formatting.
diff --git a/src/fmt/friendly/printer.rs b/src/fmt/friendly/printer.rs index f61bf4c..c2a9a07 100644 --- a/src/fmt/friendly/printer.rs +++ b/src/fmt/friendly/printer.rs @@ -1,6 +1,6 @@ use crate::{ fmt::{ - util::{DecimalFormatter, FractionalFormatter}, + util::{FractionalFormatter, IntegerFormatter...
6
99
99
1467f47e36ea69c6db079df4e7233943225a348d
Andrew Gallant
2025-12-23T01:44:21
fmt: get rid of `Decimal::end` It's not used. I believe it was being used when `Decimal` was also responsible for formatting the fractional part of a floating point number. But it isn't needed any more.
diff --git a/src/fmt/util.rs b/src/fmt/util.rs index fc0ddbf..a51e39e 100644 --- a/src/fmt/util.rs +++ b/src/fmt/util.rs @@ -83,7 +83,6 @@ impl Default for DecimalFormatter { pub(crate) struct Decimal { buf: [u8; Self::MAX_LEN as usize], start: u8, - end: u8, } impl Decimal { @@ -98,11 +97,8 @@ impl D...
1
4
8
5b27b22096a39e1b022ef362c17b777f41688600
Andrew Gallant
2025-12-23T01:02:11
fmt: simplify decimal formatter I don't think we need to keep track of maximum digits separately for signed and unsigned integer formatting. I think this might mean that signed integer formatting uses an extra byte of stack space, but, it can also have a sign. Unlike unsigned integer formatting. So no space is wasted ...
diff --git a/src/fmt/util.rs b/src/fmt/util.rs index fe214de..fc0ddbf 100644 --- a/src/fmt/util.rs +++ b/src/fmt/util.rs @@ -14,11 +14,9 @@ use crate::{ /// faster. We roll our own which is a bit slower, but gets us enough of a win /// to be satisfied with and with (almost) pure safe code. /// -/// By default, this ...
1
16
114
4d2041567c2b98f9105444cd8e16f9cb616271b6
Andrew Gallant
2025-12-22T19:54:35
binary-size: remove many uses of `write!` Using `write!` unnecessarily when a simple `f.write_str` would do ends up generating more LLVM lines on my Biff `cargo llvm-lines` benchmark. This PR replaces those uses of `write!` with `Formatter::write_str`.
diff --git a/crates/jiff-static/src/shared/posix.rs b/crates/jiff-static/src/shared/posix.rs index eb6b29e..6297171 100644 --- a/crates/jiff-static/src/shared/posix.rs +++ b/crates/jiff-static/src/shared/posix.rs @@ -1673,8 +1673,7 @@ impl core::fmt::Display for PosixJulianNoLeapError { f.write_str("in...
14
79
97
21f69521cdeba481b66ee8bf24bd065967363b04
Andrew Gallant
2025-12-22T19:15:40
shared: remove escaping and UTF-8 routines from `shared` module With the error refactor, these are no longer used. Namely, while switching to structured errors, I took that opportunity to slim down errors so that we are not repeating parts of the input as often.
diff --git a/crates/jiff-static/src/shared/util/escape.rs b/crates/jiff-static/src/shared/util/escape.rs index 5e3b8b3..e5b182b 100644 --- a/crates/jiff-static/src/shared/util/escape.rs +++ b/crates/jiff-static/src/shared/util/escape.rs @@ -122,3 +122,4 @@ impl core::fmt::Debug for RepeatByte { Ok(()) } ...
10
210
222
a50f6797ce06f78e1071acb653c4abc0f024b2c7
Andrew Gallant
2025-12-22T14:48:24
logging: small tweak to formatting
diff --git a/src/tz/db/bundled/enabled.rs b/src/tz/db/bundled/enabled.rs index 7592ea2..434b3a0 100644 --- a/src/tz/db/bundled/enabled.rs +++ b/src/tz/db/bundled/enabled.rs @@ -26,8 +26,8 @@ impl Database { Err(_err) => { warn!( "failed to parse TZif data from bundled ...
1
2
2
523b55bc1a12d8c71fae36f50a644c7570c50133
Andrew Gallant
2025-12-22T14:40:20
error: add `Error::is_invalid_parameter` predicate
diff --git a/src/error/mod.rs b/src/error/mod.rs index 52edc89..461be2b 100644 --- a/src/error/mod.rs +++ b/src/error/mod.rs @@ -111,6 +111,80 @@ impl Error { matches!(*self.root().kind(), Range(_) | SlimRange(_) | ITimeRange(_)) } + /// Returns true when this error originated as a result of an inval...
1
74
0
3cb0b6aefa01f58498ab22fa18f8e0e1fe357697
Andrew Gallant
2025-12-21T15:34:57
error: add `Error::is_crate_feature` predicate There's only a few of these, but they seem like a distinct category from other error types. It's also nice to do this and standarize on the specific error message. The other reason I wanted to do this was to distinguish it from a possible "is configuration error" categor...
diff --git a/src/error/mod.rs b/src/error/mod.rs index 14ae0c1..52edc89 100644 --- a/src/error/mod.rs +++ b/src/error/mod.rs @@ -110,6 +110,21 @@ impl Error { use self::ErrorKind::*; matches!(*self.root().kind(), Range(_) | SlimRange(_) | ITimeRange(_)) } + + /// Returns true when this error o...
6
88
23
8e8033a29d6e7178167207e5604f632dd20e95a3
Andrew Gallant
2025-12-21T14:09:15
error: add `Error::is_range` predicate I'm somewhat concerned that this doesn't cover all cases, but I think this should be a good start.
diff --git a/src/civil/date.rs b/src/civil/date.rs index 5d53c6c..e56f66f 100644 --- a/src/civil/date.rs +++ b/src/civil/date.rs @@ -1059,7 +1059,7 @@ impl Date { let nth = t::SpanWeeks::try_new("nth weekday", nth)?; if nth == C(0) { - Err(Error::from(E::NthWeekdayNonZero)) + E...
3
25
3
259e8134ef07540beec9d1ed507b9a7a95aa9ba2
Andrew Gallant
2025-12-21T13:13:11
error: implement an error chaining iterator We previously only used this in the `core::fmt::Display` trait implementation, so we just inlined it there. But we'll want to use it to get the root of the chain for error predicates, so this commit does some refactoring to provide a standalone internal iterator for the erro...
diff --git a/crates/jiff-static/src/shared/tzif.rs b/crates/jiff-static/src/shared/tzif.rs index 6406578..1b60a78 100644 --- a/crates/jiff-static/src/shared/tzif.rs +++ b/crates/jiff-static/src/shared/tzif.rs @@ -503,7 +503,6 @@ impl TzifOwned { /// and the timestamps in TZif data are, of course, all in UTC.) ...
4
88
60
831d3efb4da42bc70f7ac0294bc68a5c077bcf3d
Andrew Gallant
2025-12-20T18:35:20
shared: move itime range error into itime module This follows the pattern I used for TZif and POSIX time zone parsing.
diff --git a/crates/jiff-static/src/shared/mod.rs b/crates/jiff-static/src/shared/mod.rs index cfb528f..067be4b 100644 --- a/crates/jiff-static/src/shared/mod.rs +++ b/crates/jiff-static/src/shared/mod.rs @@ -474,7 +474,6 @@ pub struct PosixOffset { // Does not require `alloc`, but is only used when `alloc` is enabl...
9
258
221
ac0054c72f72f4ea131128200d543b0c99337e68
Andrew Gallant
2025-12-20T01:36:03
shared: replace some uses of `write!` I believe this reduces code size.
diff --git a/src/shared/posix.rs b/src/shared/posix.rs index c9a4014..648ad39 100644 --- a/src/shared/posix.rs +++ b/src/shared/posix.rs @@ -342,12 +342,11 @@ impl<ABBREV: AsRef<str> + Debug> PosixTimeZone<ABBREV> { impl<ABBREV: AsRef<str>> core::fmt::Display for PosixTimeZone<ABBREV> { fn fmt(&self, f: &mut co...
1
37
19
f2751642392ec2044754e2a122b8e23a5902b7d2
Andrew Gallant
2025-12-20T01:24:08
shared: remove `alloc` cfgs I guess parsing a POSIX time zone doesn't actually require `alloc` any more. That's cool.
diff --git a/crates/jiff-static/src/shared/posix.rs b/crates/jiff-static/src/shared/posix.rs index fef2428..1296adf 100644 --- a/crates/jiff-static/src/shared/posix.rs +++ b/crates/jiff-static/src/shared/posix.rs @@ -1577,10 +1577,8 @@ impl<'s> Parser<'s> { } } -// Tests require parsing, and parsing requires al...
4
625
255
d024322b2728181502d8c1812156348562a0fd9b
Andrew Gallant
2025-12-20T00:20:40
shared: use structured errors for TZif parsing For this one, I defined the error type inline with it. This just felt right and it does stay reasonably encapsulated. I'll probably swing back and do the same for the time range error type?
diff --git a/src/error/mod.rs b/src/error/mod.rs index 1d81b23..85a158a 100644 --- a/src/error/mod.rs +++ b/src/error/mod.rs @@ -161,6 +161,11 @@ impl Error { Error::from(ErrorKind::Shared2(err)) } + /// Creates a new error from the special TZif error type. + pub(crate) fn tzif(err: crate::shared:...
3
592
220
973876a9f1188f7026ed40dc330bc97b03b6245d
Andrew Gallant
2025-12-19T20:55:39
shared: use structured errors for `shared/itime`
diff --git a/crates/jiff-cli/cmd/generate/shared.rs b/crates/jiff-cli/cmd/generate/shared.rs index be1783e..580d89a 100644 --- a/crates/jiff-cli/cmd/generate/shared.rs +++ b/crates/jiff-cli/cmd/generate/shared.rs @@ -79,7 +79,7 @@ fn copy(srcdir: &Path, dstdir: &Path, dir: &Path) -> anyhow::Result<()> { fn copy_rust_s...
13
394
175
3a832162bed7f1e170ac92e7770908db9d51a4f1
Andrew Gallant
2025-12-19T18:11:17
fmt: some minor perf improvements to `strptime` This adds a little inlining back to help reclaim a small performance regression in `strptime`. This improves the relevant benchmark from 70ns to 64ns on my machine.
diff --git a/src/fmt/strtime/mod.rs b/src/fmt/strtime/mod.rs index 8c1915e..84e1c1c 100644 --- a/src/fmt/strtime/mod.rs +++ b/src/fmt/strtime/mod.rs @@ -1571,18 +1571,20 @@ impl BrokenDownTime { /// ``` #[inline] pub fn to_timestamp(&self) -> Result<Timestamp, Error> { - #[cold] - #[inline(...
1
30
14
3765a52b8d1e45a3cee182babc32785d4ed9d39c
Andrew Gallant
2025-11-16T18:41:20
fmt: add `&mut dyn Write` impl for `jiff::fmt::Write` I was playing with using this to avoid parametric polymorphism costs, but I couldn't find my way to a concrete benefit. Either way, this impl should exist for people that want to use it.
diff --git a/src/fmt/mod.rs b/src/fmt/mod.rs index b5886fa..e7d22b4 100644 --- a/src/fmt/mod.rs +++ b/src/fmt/mod.rs @@ -334,6 +334,17 @@ impl<W: Write> Write for &mut W { } } +impl Write for &mut dyn Write { + fn write_str(&mut self, string: &str) -> Result<(), Error> { + (**self).write_str(string) +...
1
11
0
42080c4e71fa25a16a64c5ec460f86ff777922d6
Andrew Gallant
2025-11-16T18:40:22
fmt/strtime: avoid inlining the `or_else` cases These all generally call `BrokenDownTime::to_date`, which adds a fair bit of code. Since these are the uncommon case, let's not inline that goop.
diff --git a/src/fmt/strtime/format.rs b/src/fmt/strtime/format.rs index 5d5d9a5..e5e8064 100644 --- a/src/fmt/strtime/format.rs +++ b/src/fmt/strtime/format.rs @@ -339,7 +339,10 @@ impl<'c, 'f, 't, 'w, W: Write, L: Custom> Formatter<'c, 'f, 't, 'w, W, L> { let day = self .tm .day - ...
1
99
27
40941a5188302db120a9b914e9ae73e6121efa2a
Andrew Gallant
2025-11-16T18:39:33
fmt/strtime: tighten up `to_timestamp` and `to_date` Both of these functions are marked as `inline`, but they can be quite beefy. Instead, set the inlineable implementation as the common case, and put the rest behind a non-inlineable function.
diff --git a/src/fmt/strtime/mod.rs b/src/fmt/strtime/mod.rs index 9455dd0..32d3fda 100644 --- a/src/fmt/strtime/mod.rs +++ b/src/fmt/strtime/mod.rs @@ -1584,21 +1584,27 @@ impl BrokenDownTime { /// ``` #[inline] pub fn to_timestamp(&self) -> Result<Timestamp, Error> { + #[cold] + #[inline(...
1
67
46
37803515434b2ad5f1e3c78163601d44a420949d
Andrew Gallant
2025-11-16T18:38:10
rangeint: don't try to inline error constructor
diff --git a/src/util/rangeint.rs b/src/util/rangeint.rs index a20ec34..c1a811c 100644 --- a/src/util/rangeint.rs +++ b/src/util/rangeint.rs @@ -379,7 +379,7 @@ macro_rules! define_ranged { /// dependent bounds. For example, when the day of the month is out /// of bounds. The maximum value can...
1
1
1
54c1b2d25b8b35929140b3d929f85bc26fa72429
Andrew Gallant
2025-11-14T21:28:06
error: make `map_err` closures as `cold` and non-inlineable This doesn't seem to have much impact as shown via a `cargo llvm-lines`. But this still seems like good sense to have.
diff --git a/src/error.rs b/src/error.rs index 0675547..66e9497 100644 --- a/src/error.rs +++ b/src/error.rs @@ -647,7 +647,11 @@ impl ErrorContext for Error { impl<T> ErrorContext for Result<T, Error> { #[cfg_attr(feature = "perf-inline", inline(always))] fn context(self, consequent: impl IntoError) -> Resu...
1
10
2