Repository navigation
Releases
Releases · antonsoo/logdelta
logdelta 0.3.4
cargo install logdeltaFixed
- A file that is not text was read as a log. A PNG or an archive given by mistake (a wrong artifact path in CI) became hundreds of junk templates, and the diff exited 0 or 1 like any other run. A NUL byte in the first block now stops it with
<file>: not a text file, in the CLI and in the web demo (where the editor keeps what it had). - Read errors exit 2. A missing or unreadable log exited 1, the code that means "found something", so a CI gate could not tell one from a regression. 0, 1 and 2 now mean what they mean for
diffandgrep; the README and the GitHub Actions recipe say so. - The GitHub Actions recipe downloaded a release binary that no release had ever carried, so its step failed with a 404. Releases now carry a static Linux x86-64 binary,
logdelta-x86_64-unknown-linux-musl.tar.gz, with its SHA-256 (0.3.3's was added afterwards), and the recipe fetches that.
Changed
- Gzip is recognized by its first two bytes, not by a
.gzname, in the CLI as it already was in the web demo: a compressed log saved without the extension, or piped on stdin, is read.
logdelta 0.3.3
cargo install logdeltaFixed
- A long path left the human report's locations without their file name and line numbers. Each location line is fitted to the terminal (120 columns in a pipe) by cutting its end, which is where
pytest-fail.log:14-23is; with a CI workspace or a temporary directory in front, every location read/home/runner/work/.../artifacts/lo…. The path is now cut at its start (…/artifacts/logs/pytest-fail.log:14-23). The snapshot tests run on relative paths, so they no longer fail when the repository is checked out somewhere deep.
The rest of this release is in the web demo; the package is otherwise unchanged.
Changed
- The page's fonts are served by the page itself. They came from Google Fonts, the one request the page made to another origin; the same font files (every subset, as Google serves them to a current browser) are now in
web/src/fonts/, with their SIL Open Font License texts. Nothing looks different: screenshots before and after match. The page now loads with every other host blocked.
Security
- The built page carries a Content-Security-Policy. Scripts, styles, fonts and workers load from the page's own origin only, and
connect-src 'self'has the browser refuse to send what you give the page to any other host, even for a script injected through a bug in how the page renders a file. Inline event handlers andevalare not allowed ('wasm-unsafe-eval'is: the engine is WebAssembly). Every control was exercised in Chromium and Firefox with a listener for policy violations: none.
Accessibility
- Checked with axe-core (WCAG 2.1 A and AA, and its best-practice rules) in light and dark, at desktop and phone widths, on each example: no findings now. The faint text was 3.5:1 to 3.8:1. The baseline tab list held two buttons that are not tabs (add, remove): they sit beside it now. Log excerpts that scroll sideways can take keyboard focus.
- The log editors showed keyboard focus only as their 1-pixel border turning from dark to light grey (axe-core does not check focus indicators; tabbing through every site and comparing each stop focused and unfocused does). They get the 2-pixel focus ring every other control has.
logdelta 0.3.2
cargo install logdeltaFixed
- Logs saved by Windows tools.
pytest > run.login Windows PowerShell writes UTF-16 with a byte-order mark: compared with UTF-8 baselines from CI, such a target shared no line with them and all of it was reported as new (34 templates and 9 findings for the pytest example, which has 21 and 3). A mark in front of UTF-8 made the log's first line a template of its own, one extra finding. Both encodings are recognized by the mark and read as the text they hold, in files, in.gzfiles and on stdin; the reports are then identical to those for the plain files. The web demo reads dropped files the same way.
logdelta 0.3.1
cargo install logdeltaFixed
- The minimum Rust version.
rust-versionsaid 1.80, and on 1.80cargo install logdeltastopped inside a dependency:failed to parse manifest ... clap_derive-4.6.7/Cargo.toml: feature edition2024 is required. clap has needed Rust 1.85 since 4.6.rust-versionis 1.85 now, so an older toolchain is told that in one line, and CI has a job that builds and tests on 1.85.0 with the locked dependencies. No code changed.
logdelta 0.3.0
cargo install logdeltaChecked against real CI logs for the first time: a failing pytest job of the pytest project on GitHub Actions (854 lines) and three passing runs of the same job (about 1,270 lines each), then a vitest job and a tokio job the same way. The numbers below are 0.2.3 from crates.io and this release on the same files.
Added
- Blocks. Findings whose lines sit together are reported as one. The pytest failure was 311 findings, 543 lines of report: every line of the traceback its own NEW finding, every line of the upload steps the job never reached its own GONE finding. It is now 4: the traceback (51 templates), the short test summary (27), the lines the job exited on (5), and one GONE block of 198 templates. vitest: 26 findings to 6. tokio: 99 to 13.
- A NEW block is a run of new lines, with up to two known lines let through, found while the target is read. It prints one line per template with its line number, and
×Nwhere a template has more lines. The same stack trace logged 25 times is one block that says so. - A GONE block is located in the first baseline, the run that still had the lines.
- A template folds into a block only if all of its lines are in such groups, so a message that also occurs on its own keeps its own finding and its count. Grouping never makes a report longer.
--block-lines N(default 12) sets how much of a long block is printed: its start and its end.--flatturns grouping off.--jsongains ablocksarray and ablockindex on each finding that is part of one;findingsstill lists every template. Memory stays flat: a few numbers per template, nothing per line.
- A NEW block is a run of new lines, with up to two known lines let through, found while the target is read. It prints one line per template with its line number, and
- A GONE finding shows the line the first baseline had, with its line number (
first_baseline_line_noandfirst_baseline_rawin--json), where it used to show the template alone. - Test-runner progress lines are masked:
....s....x.... [ 42%]becomes<PROGRESS> [ <NUM>%], or<PROGRESS!>when it holds anForE. Which tests land on which line depends on scheduling, so two passing runs of the pytest job diffed to 27 findings, all of them dots. Now 0. - OSC escape sequences (hyperlinks, window titles) are stripped before mining, as CSI color codes already were.
- Library:
blocks::Block,DiffResult::blocks,DiffOptions::group,scoring::count_g_test,Drain::add_token_slice,mask::strip_escapes,output::printable.
Changed
- CHANGED needs the count to move, not only the share. The score compares a template's share of the log, and a share moves whenever the run is shorter or longer for other reasons: the failing pytest job printed its 257
[new branch]lines 258 times in a log a third shorter, and that was a CHANGED finding ("more often", score 38). A second G-test on the counts alone (the baselines' total against the target's, as two Poisson counts) now has to clear the same cutoff, in the same direction. A bigger run with the same mix still changes nothing, which is what the share test is for. - A NEW VALUE on a line inside a NEW block is part of the block and is not listed again.
- Reports print log lines without the log's own escape sequences and control characters. A colored CI log put its color codes into the terminal report (76 lines of the pytest report carried them, also under
--color never), where a line cut to the terminal width left the color on for everything after it, and into Markdown tables as stray[31m.--jsonkeeps every line as it was read. - The header counts what the report shows and says what
--flatwould:4 findings (284 with --flat). - The web demo shows the engine's blocks (it grouped new lines on its own before, by first line only), has a GONE block card, cuts a long block to its start and end with a button for the rest, and strips color codes from the lines it shows.
Fixed
- A line that is mostly placeholders never matched its own template. The similarity between a line and a template skipped positions where both hold the same masking placeholder, but still divided by the full length, so
<NUM> <NUM> <HEX> okscored 1/4 against itself and every such line started a new template. 100,000 lines of that shape were 100,000 templates, 77.55 s and 45.7 MB intemplates(each line compared against every template before it); they are one template in 0.18 s and 7.2 MB. In a diff, a line identical in every run was reported NEW: the toolchain line- 1.98.1 aarch64-apple-darwin 48a229ce…in tokio's logs, between two passing runs. Such positions are now left out of both sides of the fraction.
logdelta 0.2.3
cargo install logdeltaChanged
- Reports show at most 400 characters of a template, a value or a log line, and say how much they left out. Logs do contain lines of hundreds of kilobytes (a minified bundle, a base64 payload, one JSON document), and one finding on such a line printed all of it: a 120 KB line made a 120 KB terminal report and a 240 KB Markdown table, too large for a PR comment. The same finding is now about 1 KB in either.
--jsonstill carries every line whole, and output for ordinary logs is unchanged.
logdelta 0.2.2
cargo install logdeltaAdded
- Crate-level documentation with a tested example of the library API (
analysis::diff_linesover two logs in memory), so the docs.rs page opens with more than a module list. The README has a short "As a library" section.
Fixed
- Four documentation links pointed at private items and rendered as dead links.
logdelta 0.2.1
cargo install logdeltaAdded
- Published to crates.io:
cargo install logdelta. The crate leaves out the browser demo, the screenshots and the large benchmark fixtures.