Sqlite Cache Backend by mookums · Pull Request #2631 · lightpanda-io/browser

mookums · 2026-06-03T15:36:21Z

This is a new SQLite backend for the Cache. This is built off of the Cache Revalidation PR (#2570)

krichprollsch · 2026-06-23T13:28:37Z

+            .{trimmed_dir},
+            0,
+        );
+        defer allocator.free(cache_path);


What about moving this logic into config directly? We could create something like config.httpCacheSqlitePath() returning an nullable path name? I would remove this code from Network which is not really its place and maybe allow us to have a config flag to customize this path.

krichprollsch · 2026-06-23T14:30:59Z

+                        .path = cache_path,
                        .err = e,
                    });
                    return e;


Maybe we could show an error + disable the cache in case of error to avoid process hard stop?
but I'm not sure it's a good thing 🤔

krichprollsch · 2026-06-23T14:32:43Z

+    const body_entry = try conn.row(
+        "select data from body where url = $1",
+        .{url},
+    ) orelse @panic("valid metadata must have a body");


idk about that one. I would prefer logging an error and return a null body instead.

what about assert then to get the crash report at least?

this comment has not been addressed AFAIK

krichprollsch · 2026-06-23T14:34:53Z

+
+    var lower_name: [256]u8 = undefined;
+    for (meta.headers) |h| {
+        const name = std.ascii.lowerString(lower_name[0..h.name.len], h.name);


Shouldn't we handle h.name.len > 256 ?

krichprollsch · 2026-06-23T14:35:05Z

+        );
+    }
+    for (meta.vary_headers) |h| {
+        const name = std.ascii.lowerString(lower_name[0..h.name.len], h.name);


same about h.name length

krichprollsch · 2026-06-23T14:36:26Z

    self.pool.deinit(allocator);
 }

+pub fn Migrations(comptime migrations: []const [:0]const u8) type {


Should we keep src/storage/sqlite/migrations.zig?

I'm not sure I see the reason for this to be a generic, nor what the hash provides. What case does the hash protect against? The DB getting corrupted? The developer (us) making a build-time mistake? Also, it isn't cryptographic, so whatever it does, it only does sometimes.

It might not matter for a small/simple DB schema, but I think migrations should be code, not text. There are some migrations that simply require pulling data into code.

We no longer track hash and we also support either just SQL or code as the migration.

krichprollsch · 2026-06-23T15:00:08Z

This is nice to be able to use duckdb CLI to explore the cache data :)

cache D select * from metadata where must_revalidate != 0;
┌─────────────────────────────────────────────────────────────────────────────────────────────────────────────────────┬────────┬────────────┬──────────────┬─────────┬─────────────────┬──────────────────────────────────────┬───────────────────────────────┐
│                                                         url                                                         │ status │ stored_at  │ age_at_store │ max_age │ must_revalidate │                 etag                 │         last_modified         │
│                                                       varchar                                                       │ int64  │   int64    │    int64     │  int64  │      int64      │               varchar                │            varchar            │
├─────────────────────────────────────────────────────────────────────────────────────────────────────────────────────┼────────┼────────────┼──────────────┼─────────┼─────────────────┼──────────────────────────────────────┼───────────────────────────────┤
│ https://improving.duckduckgo.com/t/page_home_tagline_impression?2117232&b=unknown&d=d&l=en&p=other                  │    200 │ 1782224467 │            0 │       0 │               1 │ NULL                                 │ Mon, 28 Sep 1970 06:00:00 GMT │
│ https://improving.duckduckgo.com/t/page_home_searchbox_ai_impression?3479694&b=unknown&d=d&l=en&p=other&atb=v539-7… │    200 │ 1782224467 │            0 │       0 │               1 │ NULL                                 │ Mon, 28 Sep 1970 06:00:00 GMT │
│ https://improving.duckduckgo.com/t/page_home_impression?8744613&theme=light&b=unknown&d=d&l=en&p=other&atb=v539-7&… │    200 │ 1782224467 │            0 │       0 │               1 │ NULL                                 │ Mon, 28 Sep 1970 06:00:00 GMT │
│ https://duckduckgo.com/t.js?q=lightpanda&l=us-en&p=1&s=0&dl=en&ct=FR&bing_market=en-US&p_ent=&dp=xXvfRl5mggeuTWh4h… │    200 │ 1782224476 │            0 │       0 │               1 │ NULL                                 │ NULL                          │
│ https://hn.algolia.com/public/main-6e634771f729331a3c3c.js                                                          │    200 │ 1782226454 │            0 │       0 │               1 │ "1755549896.0-2608202-407047307"     │ Mon, 18 Aug 2025 20:44:56 GMT │
│ https://js.driftt.com/include/1782224700000/6nmy7z65gsxc.js                                                         │    200 │ 1782224599 │            0 │       0 │               1 │ W/"d45d26b9900d0bece786bd59b23c5006" │ Thu, 26 Mar 2026 08:37:12 GMT │
│ https://config.gorgias.chat/gorgias-chat-bundle-loader.js?applicationId=15954                                       │    200 │ 1782226616 │         3537 │       0 │               1 │ W/"71a-vOzeu8cLY+IjFLWstPX9XvL10zs"  │ NULL                          │
│ https://www.googleadservices.com/pagead/conversion/971805850/?random=1782224689183&cv=11&fst=1782224689183&fmt=7&b… │    200 │ 1782224690 │            0 │       0 │               1 │ NULL                                 │ NULL                          │
│ https://config.gorgias.chat/bundle-loader/01GYCCC51FF4CA6RW7XJH50SCK?source=shopify1click&shop=kit-and-ace.myshopi… │    200 │ 1782226638 │         3557 │       0 │               1 │ W/"74d-X+O5AqWK9m1LWrn+UvfhH5u8pSY"  │ NULL                          │
│ https://www.googleadservices.com/pagead/conversion/971805850/?random=1782224718058&cv=11&fst=1782224718058&fmt=7&b… │    200 │ 1782224719 │            0 │       0 │               1 │ NULL                                 │ NULL                          │
│ https://www.googleadservices.com/pagead/conversion/971805850/?random=1782224718143&cv=11&fst=1782224718143&fmt=7&b… │    200 │ 1782224720 │            0 │       0 │               1 │ NULL                                 │ NULL                          │
│ https://www.googleadservices.com/pagead/conversion/971805850/?random=1782224755238&cv=11&fst=1782224755238&fmt=7&b… │    200 │ 1782224756 │            0 │       0 │               1 │ NULL                                 │ NULL                          │
│ https://www.googleadservices.com/pagead/conversion/971805850/?random=1782224755314&cv=11&fst=1782224755314&fmt=7&b… │    200 │ 1782224756 │            0 │       0 │               1 │ NULL                                 │ NULL                          │
│ https://improving.duckduckgo.com/t/page_home_tagline_impression?9740447&b=unknown&d=d&l=en&p=other                  │    200 │ 1782225829 │            0 │       0 │               1 │ NULL                                 │ Mon, 28 Sep 1970 06:00:00 GMT │
│ https://improving.duckduckgo.com/t/page_home_searchbox_ai_impression?5392774&b=unknown&d=d&l=en&p=other&atb=v539-7… │    200 │ 1782225829 │            0 │       0 │               1 │ NULL                                 │ Mon, 28 Sep 1970 06:00:00 GMT │
│ https://improving.duckduckgo.com/t/page_home_impression?1190434&theme=light&b=unknown&d=d&l=en&p=other&atb=v539-7&… │    200 │ 1782225829 │            0 │       0 │               1 │ NULL                                 │ Mon, 28 Sep 1970 06:00:00 GMT │
│ https://improving.duckduckgo.com/t/page_home_tagline_impression?8429873&b=unknown&d=d&l=en&p=other                  │    200 │ 1782226412 │            0 │       0 │               1 │ NULL                                 │ Mon, 28 Sep 1970 06:00:00 GMT │
│ https://improving.duckduckgo.com/t/page_home_searchbox_ai_impression?9690011&b=unknown&d=d&l=en&p=other&atb=v539-7… │    200 │ 1782226412 │            0 │       0 │               1 │ NULL                                 │ Mon, 28 Sep 1970 06:00:00 GMT │
│ https://duckduckgo.com                                                                                              │    200 │ 1782226421 │            0 │       0 │               1 │ "6a3a9d36-9c21"                      │ NULL                          │
│ https://improving.duckduckgo.com/t/page_home_tagline_impression?7683633&b=unknown&d=d&l=en&p=other                  │    200 │ 1782226424 │            0 │       0 │               1 │ NULL                                 │ Mon, 28 Sep 1970 06:00:00 GMT │
│ https://improving.duckduckgo.com/t/page_home_searchbox_ai_impression?8002091&b=unknown&d=d&l=en&p=other&atb=v539-7… │    200 │ 1782226424 │            0 │       0 │               1 │ NULL                                 │ Mon, 28 Sep 1970 06:00:00 GMT │
│ https://improving.duckduckgo.com/t/page_home_impression?585745&theme=light&b=unknown&d=d&l=en&p=other&atb=v539-7&p… │    200 │ 1782226424 │            0 │       0 │               1 │ NULL                                 │ Mon, 28 Sep 1970 06:00:00 GMT │
│ https://duckduckgo.com/ac/?q=light&kl=wt-wt&vertical=homepage                                                       │    200 │ 1782226427 │            0 │       0 │               1 │ NULL                                 │ NULL                          │
│ https://duckduckgo.com/ac/?q=lightp&kl=wt-wt&vertical=homepage                                                      │    200 │ 1782226427 │            0 │       0 │               1 │ NULL                                 │ NULL                          │
│ https://duckduckgo.com/ac/?q=lightpa&kl=wt-wt&vertical=homepage                                                     │    200 │ 1782226428 │            0 │       0 │               1 │ NULL                                 │ NULL                          │
│ https://duckduckgo.com/ac/?q=lightpan&kl=wt-wt&vertical=homepage                                                    │    200 │ 1782226428 │            0 │       0 │               1 │ NULL                                 │ NULL                          │
│ https://duckduckgo.com/ac/?q=lightpand&kl=wt-wt&vertical=homepage                                                   │    200 │ 1782226429 │            0 │       0 │               1 │ NULL                                 │ NULL                          │
│ https://duckduckgo.com/ac/?q=lightpanda&kl=wt-wt&vertical=homepage                                                  │    200 │ 1782226429 │            0 │       0 │               1 │ NULL                                 │ NULL                          │
│ https://duckduckgo.com/t.js?q=lightpanda&l=us-en&s=0&dl=en&ct=FR&bing_market=en-US&p_ent=&ex=-1&dp=J9gRog2IaKXlQqQ… │    200 │ 1782226432 │            0 │       0 │               1 │ NULL                                 │ NULL                          │
│ https://duckduckgo.com/post3.html                                                                                   │    200 │ 1782226432 │            0 │       0 │               1 │ "6a3a9d69-88"                        │ NULL                          │
│ https://www.amazon.com/s/ref=nb_sb_noss?url=search-alias%3Daps&field-keywords=ces+jours+qui+disparaissent           │    200 │ 1782226530 │            0 │       0 │               1 │ NULL                                 │ NULL                          │
│ https://js.driftt.com/include/1782226800000/6nmy7z65gsxc.js                                                         │    200 │ 1782226539 │            0 │       0 │               1 │ W/"d45d26b9900d0bece786bd59b23c5006" │ Thu, 26 Mar 2026 08:37:12 GMT │
│ https://www.googleadservices.com/pagead/conversion/971805850/?random=1782226623304&cv=11&fst=1782226623304&fmt=7&b… │    200 │ 1782226624 │            0 │       0 │               1 │ NULL                                 │ NULL                          │
└─────────────────────────────────────────────────────────────────────────────────────────────────────────────────────┴────────┴────────────┴──────────────┴─────────┴─────────────────┴──────────────────────────────────────┴───────────────────────────────┘
  33 rows

krichprollsch · 2026-06-23T15:03:32Z

+
+const SqliteCacheMigrations = Migrations(&.{
+    \\ create table metadata (
+    \\      url               text not null primary key,


I'm wondering if using the same hash as migration as primary key wouldn't be better?

karlseguin · 2026-06-24T02:45:48Z


-    const cache = if (config.httpCacheDir()) |cache_dir_path|
-        Cache{
+    const cache = if (config.httpCacheDir()) |cache_dir_path| blk: {


I don't think there should be this much caching logic in Network.zig.

karlseguin · 2026-06-24T03:05:12Z

    self.pool.deinit(allocator);
 }

+pub fn Migrations(comptime migrations: []const [:0]const u8) type {


I'm not sure I see the reason for this to be a generic, nor what the hash provides. What case does the hash protect against? The DB getting corrupted? The developer (us) making a build-time mistake? Also, it isn't cryptographic, so whatever it does, it only does sometimes.

It might not matter for a small/simple DB schema, but I think migrations should be code, not text. There are some migrations that simply require pulling data into code.

karlseguin · 2026-06-24T03:27:54Z

+    \\      foreign key (url) references metadata(url) on delete cascade
+    \\ )
+    ,
+    \\ create table header (


what about:

create table cache ( url text not null primary key, vary bool not null, response text not null, body blob not null )

Only url and varyare queried. response can be a JSON representation of the cache. body is extracted from response to avoid the base64 bloating of any binary data.

krichprollsch · 2026-07-02T13:24:30Z

If we want to cleanup the db at some point, for example to reduce its size, do we have an SQL request to do it easily?

krichprollsch · 2026-07-02T13:25:26Z

I had a crash with integration tests + warmup cache (and a rebased branch)

Segmentation fault at address 0x77bc847ab930
/home/pierre/wrk/browser-claude/src/network/cache/Cache.zig:218:42: 0x47343c9 in deinit (lightpanda)
    pub fn deinit(self: CachedData) void {
                                         ^
/home/pierre/wrk/browser-claude/src/network/layer/CacheLayer.zig:203:29: 0x481d669 in serveFromCache (lightpanda)
    defer cached.data.deinit();
                            ^
/home/pierre/wrk/browser-claude/src/network/layer/CacheLayer.zig:84:35: 0x4733cee in run (lightpanda)
                    serveFromCache(t, c) catch |err| {
                                  ^
/home/pierre/wrk/browser-claude/src/browser/HttpClient.zig:508:16: 0x38218ff in drainNextTickQueue (lightpanda)
        ntn.run(transfer, ntn.ctx);
               ^
/home/pierre/wrk/browser-claude/src/browser/HttpClient.zig:456:32: 0x38229ab in tick (lightpanda)
    try self.drainNextTickQueue();
                               ^
/home/pierre/wrk/browser-claude/src/browser/Runner.zig:302:29: 0x44afd00 in _tick__anon_516032 (lightpanda)
        try http_client.tick(@intCast(ms_to_wait), .all);
                            ^
/home/pierre/wrk/browser-claude/src/browser/Runner.zig:150:39: 0x44b00a2 in _wait__anon_516026 (lightpanda)
        const tick_result = self._tick(is_cdp, 200, conditions) catch |err| {
                                      ^
/home/pierre/wrk/browser-claude/src/browser/Runner.zig:75:23: 0x44b050b in waitForFrameCDP (lightpanda)
    _ = try self._wait(true, timeout_ms, &conditions);
                      ^
/home/pierre/wrk/browser-claude/src/cdp/CDP.zig:269:34: 0x44b0685 in pageWait (lightpanda)
    return runner.waitForFrameCDP(page.frame_id, ms, .done);
                                 ^
/home/pierre/wrk/browser-claude/src/cdp/CDP.zig:246:18: 0x44b077b in tick (lightpanda)
    self.pageWait(wait_ms) catch |wait_err| switch (wait_err) {
                 ^
/home/pierre/wrk/browser-claude/src/Server.zig:259:30: 0x435828d in handleConnection (lightpanda)
        const next = cdp.tick() catch |err| {
                             ^
/usr/local/zig-0.15.2/lib/std/Thread.zig:509:13: 0x41f4890 in callFn__anon_441681 (lightpanda)
            @call(.auto, f, args);
            ^
/usr/local/zig-0.15.2/lib/std/Thread.zig:781:30: 0x40e1cc3 in entryFn (lightpanda)
                return callFn(f, args_ptr.*);
                             ^
./nptl/pthread_create.c:447:8: 0x77bd0d69caa3 in start_thread (pthread_create.c)
../sysdeps/unix/sysv/linux/x86_64/clone3.S:78:0: 0x77bd0d729c6b in __clone3 (../sysdeps/unix/sysv/linux/x86_64/clone3.S)
???:?:?: 0x0 in ??? (???)
run
└─ run exe lightpanda failure
error: the following command terminated unexpectedly:
./.zig-cache/o/cdeffea3ddf78e9d9796e5606def3d0f/lightpanda serve --http-cache-dir /tmp/lp-cache

Build Summary: 30/32 steps succeeded; 1 failed
run transitive failure
└─ run exe lightpanda failure

error: the following build command failed with exit code 1:
.zig-cache/o/d20cc142e18b434fc8b7b6f062898d2d/build /usr/local/zig-0.15.2/zig /usr/local/zig-0.15.2/lib /home/pierre/wrk/browser-claude .zig-cache /home/pierre/.cache/zig --seed 0x8ae37990 -Zc4c25bdb8f119cba run -- serve --http-cache-dir /tmp/lp-cache

krichprollsch · 2026-07-20T15:07:09Z

+    const body_entry = try conn.row(
+        "select data from body where url = $1",
+        .{url},
+    ) orelse @panic("valid metadata must have a body");


this comment has not been addressed AFAIK

krichprollsch · 2026-07-20T15:07:27Z

+    \\      name              text not null,
+    \\      value             text not null,
+    \\      vary              bool not null,
+    \\      primary key (url, name),


using url + header name as PK is a problem for response containing the same header multiple time.

krichprollsch · 2026-07-20T15:14:19Z

    const req_headers = try iter.collect(arena);

-    const cached = cache.get(arena, .{
+    const cached = try cache.get(arena, .{


You should maybe catch the error and return a missing cache + a log instead of propagating the error.

We catch where it would be missing and handle that separately within the SqliteCache impll. The error being propagated here would be something going wrong with Sqlite.

Which still isn't a reason to fail the request

krichprollsch · 2026-07-20T15:25:50Z

+
+fn insertMetadata(conn: Conn, meta: CachedMetadata, body: []const u8) !void {
+    try conn.exec(
+        \\ insert or replace into metadata


Claude complaints the FK w/ headers depends on pragma foreign_keys=on conn:

4. put correctness rests on an implicit REPLACE→FK-cascade interaction. insert or replace into metadata (:173) only clears the old body/header rows if pragma foreign_keys=on took effect on that connection; otherwise the plain insert into body (:188) hits a PK conflict. Make it explicit — delete from metadata where url = $1 first — rather than depending on a subtle corner of SQLite semantics.

karlseguin · 2026-07-21T03:39:20Z

On an expired cache hit, we pull the meta data (dupe from sqlite), pull the body (dupe from sqlite), then detect that it's expired and free everything. Because it's an arena, this may or may not be freed, so in the end we're 2x peak memory (the expired body we discarded + the new body we fetch).

This is just the most extreme example, but there are others:
1 - Same applies to header as applies to body
2 - No need to collect the request headers if the URL isn't even in the DB
3 - I don' think there's ever a need to dupe the metadata
4 - renew in particular doesn't need to fetch the metadata (with duping) just to update the row

I think there's a few more. There's a hint of more tightly coupled and more efficient code with the early exit around 'vary mismatch', but even this has duped every header by the time it's checked.

Aside from that, there's still some unused .file references and renew does a select, followed by update within the same transaction. This is a recipe for an BUSY error. Should probably use begin immediate

karlseguin · 2026-07-23T03:26:30Z

    const req_headers = try iter.collect(arena);

-    const cached = cache.get(arena, .{
+    const cached = try cache.get(arena, .{


Which still isn't a reason to fail the request

karlseguin · 2026-07-23T03:34:02Z

+    const conn = try self.pool.acquire();
+    defer self.pool.release(conn);
+
+    try conn.begin(.immediate);


Serializes all all cache access. In renew, it made sense because you always turned it into an write (and, ideally, writes are less common than reads). Here, the write is the edge-case. It's better to commit the expired read, and then issue the write under a distinct transaction. You can guard against another thread updating the row by changing the delete to included stored_at = $2 and passing the read stored_at.

Just split it back to how it was where get is meant just to get and it now returns an enum where we evict after if needed.

karlseguin · 2026-07-23T03:40:06Z

+        "select data from body where url = $1",
+        .{req.url},
+    ) orelse {
+        log.debug(.cache, "miss", .{ .url = req.url, .reason = "missing body " });


But the metadata row remains. This makes my previous comment about using a more concurrent-friendly transaction more complicated to implement (since deleting the metadata here would introduce a 2nd (but still very infrequent, write path). Still possible to clean this up AND have tighter transactions, but not if that end result ends up too spaghetti.

Merged the body into the other row so they're just colocated together now. This should prevent this issue of one existing while the other doesn't.

mookums · 2026-07-23T20:02:44Z

Will fail integration tests until #3044 is merged.

Huge max-age/Age header values previously either panicked in the i64 casts when storing metadata, or were silently truncated to their first 8 digits by the fixed-size lowercase buffer in CacheControl.parse. Cap them at 2^31 as RFC 9111 §1.2.2 prescribes and compare directives case-insensitively without truncation.

mookums force-pushed the sqlite-cache branch from df9d1dd to e07247c Compare June 8, 2026 15:23

mookums force-pushed the sqlite-cache branch from e07247c to 037fbb7 Compare June 18, 2026 15:39

mookums marked this pull request as ready for review June 22, 2026 14:14

krichprollsch reviewed Jun 23, 2026

View reviewed changes

karlseguin reviewed Jun 24, 2026

View reviewed changes

mookums force-pushed the sqlite-cache branch from 6b87df7 to 834a53f Compare June 24, 2026 14:17

mookums force-pushed the sqlite-cache branch 2 times, most recently from 8413ec9 to 470f713 Compare July 3, 2026 15:12

mookums force-pushed the sqlite-cache branch 5 times, most recently from 09b854f to fffa038 Compare July 20, 2026 13:44

krichprollsch reviewed Jul 20, 2026

View reviewed changes

mookums force-pushed the sqlite-cache branch from b7ed0d4 to 6c605b3 Compare July 20, 2026 21:10

karlseguin reviewed Jul 23, 2026

View reviewed changes

mookums force-pushed the sqlite-cache branch 2 times, most recently from a28a026 to 246aa9e Compare July 23, 2026 19:33

mookums force-pushed the sqlite-cache branch from bd6d1be to 25f4905 Compare July 24, 2026 13:59

mookums and others added 5 commits July 31, 2026 16:27

add SqliteCache as default local impl

7c2dd1d

use strict tables

6590546

remove primary key on url name

afcad7a

remove Storage and FsCache

4514cdb

mookums added 16 commits July 31, 2026 16:27

loadBody can return null

5a31359

manually delete and reinsert

020d1e7

get rid of file variant in CachedData

3bc2fe4

check vary before we dupe in SqliteCache.get

48bca6d

skip body dupe on expired without validators on Cache.get

fab432e

early exit on SqliteCache.get on expired without validators

741594e

always defer conn.rollback after tx begin

944ba10

add test for no validators behavior

dadc38d

remove cache eviction path from HttpClient

8d54555

add transaction begin types

bc42f88

rework get to be strictly deferred

f7fb1f0

clear old non-Vary headers on renew

c81a7c9

use CachePutRequest for puts instead of CacheMetadata

6df4c05

add format for SqliteCachePath

55d69a2

use real seconds value for migrations applied_at

4f11962

add content_type column in Cache

06e1f5d

mookums force-pushed the sqlite-cache branch from 25f4905 to 06e1f5d Compare July 31, 2026 23:42

fix minor compilation issues

af14426

Uh oh!

Conversation

mookums commented Jun 3, 2026

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

krichprollsch commented Jun 23, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

krichprollsch commented Jul 2, 2026

Uh oh!

krichprollsch commented Jul 2, 2026

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

karlseguin commented Jul 21, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

mookums commented Jul 23, 2026

Uh oh!

Reviewers

Assignees

Labels

krichprollsch commented Jun 23, 2026 •

edited

Loading

karlseguin commented Jul 21, 2026 •

edited

Loading