Closed dscho closed 1 month ago
Hrm. The tests pass for me on Windows... These failures all seem to be variations of this symptom:
Step failed: \\?\D:\a\pagefind\pagefind\pagefind\features\anchors.feature:58:9
Captured output:
Step panicked. Captured output: called `Result::unwrap()` on an `Err` value: Timeout
However, the output does not suggest that anything timed out?
Civilization {
tmp_dir: Some(
TempDir {
path: "C:\\Users\\RUNNER~1\\AppData\\Local\\Temp\\.tmp6ItJKt",
},
),
last_command_output: Some(
CommandOutput {
stdout: "\nRunning Pagefind v0.0.0 (Extended)\nRunning from: \"C:\\\\Users\\\\RUNNER~1\\\\AppData\\\\Local\\\\Temp\\\\.tmp6ItJKt\"\nSource: \"public\"\nOutput: \"public\\\\pagefind\"\n\n[Walking source directory]\nFound 4 files matching **/*.{html}\n\n[Parsing files]\nFound a data-pagefind-body element on the site.\n↳ Ignoring pages without this tag.\n\n[Reading languages]\nDiscovered 1 language: en\n\n[Building search indexes]\nTotal: \n Indexed 1 language\n Indexed 3 pages\n Indexed 49 words\n Indexed 0 filters\n Indexed 0 sorts\n\nFinished in 0.028 seconds\n",
stderr: "",
},
),
browser: None,
assigned_server_port: Some(
18145,
),
threads: [
JoinHandle {
id: Id(
5,
),
},
],
handles: [
ServerHandle {
cmd_tx: UnboundedSender {
chan: Tx {
inner: Chan {
tx: Tx {
block_tail: 0x000001a38f51fa10,
tail_position: 1,
},
semaphore: Semaphore(
1,
),
rx_waker: AtomicWaker,
tx_count: 1,
rx_fields: "...",
},
},
},
},
],
env_vars: {
"PAGEFIND_SITE": "public",
},
}
I also notice that main
fails in the same way. @bglw any ideas?
Yeah that won’t be you — the test suite is a little flaky right now, especially on the windows runners lately. I have a new testing framework on the way which should resolve that.
I’ll give the tests some pokes before I merge, which will be next week as I’m out for the weekend.
(Change looks good by the way, thank you! 🙏)
I’m out for the weekend.
On a Wednesday? CloudCannon must be a nice place to work at, do you have openings? 😁
Yeah that won’t be you — the test suite is a little flaky right now, especially on the windows runners lately. I have a new testing framework on the way which should resolve that.
Great!
Sorry for the delay! My test suite has caught up with me (and/or GitHub is only giving me the runt-of-the-litter machines 😓)
I'll run the tests locally on my mac and merge tomorrow.
@dscho this is now released on 1.1.1-alpha.1
if you're fine using the prerelease for a while :)
Pagefind has code to strip a trailing
index.html
from a URL. The intention is obviously to avoid showing/folder/index.html
when/folder/
would already do.However, the code is a bit lax about checking that
index.html
is the full file name: It would also strip that suffix for, say,/folder/my_index.html
. That, however, leads to a broken link...I noticed this because I want to convert https://git-scm.com/ to use Pagefind, but any search hitting
/docs/git-checkout-index.html
would mistakenly link to/docs/git-checkout-
.