@node-projects/speedscope 2.0.0 → 2.2.0
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- package/README.md +83 -4
- package/bin/cli.mjs +0 -0
- package/dist/release/file-format-schema.json +6 -0
- package/dist/release/index.html +2 -1
- package/dist/release/re2js.LICENSE +21 -0
- package/dist/release/release.txt +3 -3
- package/dist/release/speedscope-5DZPH3NW.js +244 -0
- package/dist/release/streamparser-json.LICENSE +21 -0
- package/docs/IMPORT_FORMATS.md +174 -0
- package/docs/JAVA_THREAD_DUMPS.md +37 -0
- package/docs/JAVA_THREAD_DUMP_ZIP.md +65 -0
- package/docs/LARGE_TRACE_PERFORMANCE.md +330 -6
- package/docs/OFFLINE.md +58 -0
- package/docs/OTEL_PROFILES.md +117 -0
- package/docs/PERF_IMPORT.md +66 -0
- package/docs/PPROF_IMPORT.md +50 -0
- package/docs/PROFILE_COMPARISON.md +118 -0
- package/docs/PROFILE_FILTERS.md +73 -0
- package/docs/PROPERTY_TESTS.md +36 -0
- package/docs/PYTHON_CPROFILE.md +25 -0
- package/docs/RELEASING.md +51 -9
- package/docs/SEARCH_AND_LAYOUT.md +385 -0
- package/docs/SINGLE_FILE.md +55 -0
- package/docs/SOURCE_AND_LABELS.md +184 -0
- package/package.json +23 -3
- package/src/lib/file-format-spec.ts +4 -0
- package/dist/release/favicon-16x16-V2DMIAZS.js.map +0 -7
- package/dist/release/favicon-32x32-THY3JDJL.js.map +0 -7
- package/dist/release/favicon-M34RF7BI.js.map +0 -7
- package/dist/release/speedscope-GHPHNKXC.css.map +0 -7
- package/dist/release/speedscope-IT6IHU5E.js +0 -188
- package/dist/release/speedscope-IT6IHU5E.js.map +0 -7
|
@@ -0,0 +1,21 @@
|
|
|
1
|
+
MIT License
|
|
2
|
+
|
|
3
|
+
Copyright (c) 2020 Juanjo Diaz
|
|
4
|
+
|
|
5
|
+
Permission is hereby granted, free of charge, to any person obtaining a copy
|
|
6
|
+
of this software and associated documentation files (the "Software"), to deal
|
|
7
|
+
in the Software without restriction, including without limitation the rights
|
|
8
|
+
to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
|
|
9
|
+
copies of the Software, and to permit persons to whom the Software is
|
|
10
|
+
furnished to do so, subject to the following conditions:
|
|
11
|
+
|
|
12
|
+
The above copyright notice and this permission notice shall be included in all
|
|
13
|
+
copies or substantial portions of the Software.
|
|
14
|
+
|
|
15
|
+
THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
|
|
16
|
+
IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
|
|
17
|
+
FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
|
|
18
|
+
AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
|
|
19
|
+
LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
|
|
20
|
+
OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
|
|
21
|
+
SOFTWARE.
|
|
@@ -0,0 +1,174 @@
|
|
|
1
|
+
# Import formats and background processing
|
|
2
|
+
|
|
3
|
+
The [README](../README.md#supported-file-formats) lists supported sources.
|
|
4
|
+
This guide describes additional fork-specific behavior and its limits.
|
|
5
|
+
|
|
6
|
+
## Perf clock weights
|
|
7
|
+
|
|
8
|
+
Explicit `perf script` periods now determine aggregate weights. CPU/task-clock
|
|
9
|
+
periods are nanoseconds of sampled CPU time; off-CPU periods remain separate
|
|
10
|
+
sleep-time profiles, and cycles remain raw counts. Missing periods retain
|
|
11
|
+
explicitly labeled inference or sample counts. Timestamp gaps are not claimed
|
|
12
|
+
as measured execution. See [perf semantics and the public regression](PERF_IMPORT.md).
|
|
13
|
+
|
|
14
|
+
## OpenTelemetry Profiles
|
|
15
|
+
|
|
16
|
+
Current OTLP Profiles JSON imports sampled stacks through a shared dictionary,
|
|
17
|
+
preserving metric weights, inline chains and thread/resource separation.
|
|
18
|
+
Timestamps do not become durations, and sample periods do not multiply values.
|
|
19
|
+
The importer currently parses the JSON in memory and creates aggregate profiles;
|
|
20
|
+
binary protobuf and older development layouts remain unsupported. See the
|
|
21
|
+
[pinned schema, limits and exact supported scope](OTEL_PROFILES.md).
|
|
22
|
+
|
|
23
|
+
## Classic Android method traces
|
|
24
|
+
|
|
25
|
+
Open the classic concatenated key/data format emitted by Android method tracing:
|
|
26
|
+
a text key section followed by a little-endian `SLOW` binary section. Versions
|
|
27
|
+
1, 2 and 3 are supported, including version-3 record padding. The file can be
|
|
28
|
+
named `.trace` or renamed; detection uses the format signature.
|
|
29
|
+
|
|
30
|
+
The importer preserves method signatures, thread identities and available Java
|
|
31
|
+
source locations. Clock metadata is required. `thread-cpu` and `wall` clocks
|
|
32
|
+
remain distinct; version-3 `dual` input creates separate labeled CPU and wall
|
|
33
|
+
profiles for each thread. Version-1 `global` means wall time. All values use
|
|
34
|
+
microseconds. CPU clocks are local to their thread; wall offsets remain relative
|
|
35
|
+
to the trace origin, so leading unobserved time can appear as idle weight.
|
|
36
|
+
The absolute date in the binary header is not added to those offsets.
|
|
37
|
+
|
|
38
|
+
Initial method exits reconstruct calls already active at trace start. Calls
|
|
39
|
+
still open at EOF close at that thread's last observed event. These are clipped
|
|
40
|
+
observation windows and do not invent work before or after the recording.
|
|
41
|
+
|
|
42
|
+
Fixed records are scanned without retaining a second parsed-record array.
|
|
43
|
+
Temporary state includes key dictionaries and per-thread stacks, alongside the
|
|
44
|
+
output builders. Key text is decoded line by line with a 16 MiB section limit
|
|
45
|
+
and 1 MiB line limit. The input buffer and final profiles still occupy memory.
|
|
46
|
+
|
|
47
|
+
Raw binary-only/streaming traces, newer ART packet formats, Perfetto protobuf,
|
|
48
|
+
Android systrace HTML and wrapped timestamp recovery are unsupported. Unknown
|
|
49
|
+
clocks, partial records, invalid method actions and decreasing per-thread clocks
|
|
50
|
+
produce explicit errors. See the
|
|
51
|
+
[synthetic fixture, generation script and AOSP references](../sample/profiles/android-method-trace/README.md)
|
|
52
|
+
for the exact supported layouts and clock policy.
|
|
53
|
+
|
|
54
|
+
## Java thread-dump archives
|
|
55
|
+
|
|
56
|
+
Traditional HotSpot `jstack` / `Thread.print` text can be opened directly or as
|
|
57
|
+
an archive of captures from one JVM. Observed stacks contribute raw samples;
|
|
58
|
+
they do not measure elapsed time. The ZIP importer checks declared sizes and
|
|
59
|
+
actual expansion, feeds bounded UTF-8 lines into the parser and supports
|
|
60
|
+
cancellation without returning partial results.
|
|
61
|
+
|
|
62
|
+
See [text snapshot semantics](JAVA_THREAD_DUMPS.md) and the
|
|
63
|
+
[ZIP format and resource policy](JAVA_THREAD_DUMP_ZIP.md). Archive support is
|
|
64
|
+
limited to classic ZIP32 STORE/DEFLATE; it is not a generic archive extractor.
|
|
65
|
+
|
|
66
|
+
## pprof metric and source identity
|
|
67
|
+
|
|
68
|
+
pprof's `default_sample_type` selects a sample metric by its string-table
|
|
69
|
+
identity. Protobuf integer objects are compared by value. If the preferred
|
|
70
|
+
metric is absent or cannot be found, the existing last-metric fallback applies.
|
|
71
|
+
For example, a recording can prefer allocation counts even when a later sample
|
|
72
|
+
column measures bytes; the displayed unit follows the chosen metric.
|
|
73
|
+
|
|
74
|
+
Function IDs and location IDs retain their complete unsigned 64-bit identity.
|
|
75
|
+
They are not truncated to their low 32 bits or rounded through a JavaScript
|
|
76
|
+
number, so IDs that share a low word remain distinct. String-table positions
|
|
77
|
+
must fit safe nonnegative integers. Source lines must fit safe positive integers.
|
|
78
|
+
|
|
79
|
+
Each location retains its own source line, falling back to the function's
|
|
80
|
+
definition line when needed. One location no longer overwrites the shared
|
|
81
|
+
function metadata used by another location. The current importer retains the
|
|
82
|
+
outermost entry of an inline-location chain; expanding all inlined functions
|
|
83
|
+
is outside this change.
|
|
84
|
+
|
|
85
|
+
Negative Go heap delta weights now import as separate negative-magnitude and
|
|
86
|
+
positive-contribution profiles. Compare positive minus negative in absolute mode
|
|
87
|
+
to inspect net changes; see [signed pprof semantics and limits](PPROF_IMPORT.md).
|
|
88
|
+
This does not add arbitrary-precision weights, a metric chooser or full inline
|
|
89
|
+
chains. See the [pprof schema used by the importer](../src/import/profile.proto).
|
|
90
|
+
|
|
91
|
+
## Background import for large recordings
|
|
92
|
+
|
|
93
|
+
### Streaming perf and folded stacks
|
|
94
|
+
|
|
95
|
+
UTF-8 folded stacks and `perf script` text now stream plain or gzip input into
|
|
96
|
+
the same incremental parsers used by ordinary imports. Supported suffixes are
|
|
97
|
+
`.folded`, `.collapsed`, `.stacks`, `.collapsedstack.txt`, `.perf`, `.perf.txt`
|
|
98
|
+
and `.linux-perf.txt`, optionally followed by `.gz`. Ordinary `.txt` files can
|
|
99
|
+
be detected as either format. Other names retain generic format detection.
|
|
100
|
+
|
|
101
|
+
The reader does not retain the complete expanded byte buffer, decoded text or
|
|
102
|
+
an array of all lines. It retains one bounded line and, for perf, one bounded
|
|
103
|
+
event. Input limits are 1 GiB expanded bytes, 1 Mi UTF-16 code units per line,
|
|
104
|
+
4 Mi code units per perf event, 65,536 stack frames and two million samples.
|
|
105
|
+
A limit failure stops import; it does not retry through an unbounded reader.
|
|
106
|
+
Malformed weights and existing perf ordering checks remain enforced.
|
|
107
|
+
|
|
108
|
+
This path also runs for small compressed inputs and when workers are unavailable.
|
|
109
|
+
It yields during parsing, exposes progress and supports cancellation. Gzip needs
|
|
110
|
+
native `DecompressionStream`; unavailable native decompression, UTF-16 input and
|
|
111
|
+
other recognized formats retain the generic fallback. Final trees, sample lists,
|
|
112
|
+
symbol processing and worker snapshots still scale with the recording.
|
|
113
|
+
|
|
114
|
+
### Worker routing and fallback
|
|
115
|
+
|
|
116
|
+
Inputs of at least **1 MiB** use a Web Worker when available, regardless of
|
|
117
|
+
filename extension. This includes perf, folded stacks, pprof, JFR, Android
|
|
118
|
+
method traces, supported ZIP archives and renamed captures. Format detection,
|
|
119
|
+
generic parsing, demangling and profile snapshot preparation happen away from
|
|
120
|
+
the main UI thread. Cancelling the task terminates its worker.
|
|
121
|
+
|
|
122
|
+
Recognized gzip perf/folded/text suffixes prefer a worker at any compressed
|
|
123
|
+
size: a tiny compressed file can expand into a large recording. Small plain
|
|
124
|
+
text stays on the main thread. The bounded text reader yields for cancellation
|
|
125
|
+
when a worker is blocked or unavailable; native gzip streaming requires
|
|
126
|
+
`DecompressionStream`. Unsupported encodings and unavailable streaming APIs use
|
|
127
|
+
the existing generic fallback, which can materialize the full input.
|
|
128
|
+
|
|
129
|
+
The worker first attempts the duration-trace and supported text streaming paths,
|
|
130
|
+
then uses the generic importer when those formats do not apply. Generic fallback
|
|
131
|
+
formats still read/materialize their input and retain their final model; routing them
|
|
132
|
+
through a worker does not make them streaming or out-of-core. Snapshot transfer,
|
|
133
|
+
geometry and final view construction still have memory costs proportional to
|
|
134
|
+
the recording. [Large-trace guidance](LARGE_TRACE_PERFORMANCE.md) describes the
|
|
135
|
+
duration-trace limits and measurements separately.
|
|
136
|
+
|
|
137
|
+
If workers are unavailable or blocked by a host's CSP, the app falls back to
|
|
138
|
+
the main-thread importer. DOM-dependent formats can also fall back after worker
|
|
139
|
+
detection. A transferred input buffer is returned before this fallback, so the
|
|
140
|
+
main thread does not attempt to parse an empty detached buffer. Synchronous
|
|
141
|
+
generic fallback parsing can block the UI until it returns; the streaming text
|
|
142
|
+
and asynchronous ZIP importers yield and check cancellation on the main thread.
|
|
143
|
+
|
|
144
|
+
Oversized supported ZIP files receive a four-byte signature preflight before
|
|
145
|
+
their complete body is read, both in the main importer and worker path. Other
|
|
146
|
+
formats retain their existing input/resource policies; the ZIP compressed-size
|
|
147
|
+
limit is not a blanket cap on every supported recording.
|
|
148
|
+
|
|
149
|
+
### Text-stream measurements
|
|
150
|
+
|
|
151
|
+
Three fresh Node 24.19.0 ARM64 processes per mode imported a synthetic file
|
|
152
|
+
containing one million copies of `root;worker;leaf 1`: 19,000,000 plain bytes
|
|
153
|
+
or 46,140 gzip bytes. All modes produced one profile with total weight 1,000,000.
|
|
154
|
+
The final model is deliberately tiny, which isolates temporary input costs;
|
|
155
|
+
these figures do not predict memory for a million distinct stacks.
|
|
156
|
+
|
|
157
|
+
| Mode | Plain median time / peak RSS | Gzip median time / peak RSS |
|
|
158
|
+
| --- | ---: | ---: |
|
|
159
|
+
| Generic materializing importer | 466 ms / 110 MiB | 853 ms / 300 MiB |
|
|
160
|
+
| Streaming with worker-style scheduling | 573 ms / 86 MiB | 629 ms / 66 MiB |
|
|
161
|
+
| Streaming with main-thread cooperative yields | 993 ms / 85 MiB | 1,092 ms / 66 MiB |
|
|
162
|
+
|
|
163
|
+
Peak RSS columns are medians of each fresh process's peak, not retained model
|
|
164
|
+
heap. Worker-style measurements skip unnecessary timer yields, as the worker
|
|
165
|
+
can be terminated externally. They time the parsing function over a Node file
|
|
166
|
+
stream, excluding actual worker startup, demangling, snapshot transfer and view
|
|
167
|
+
construction. The cooperative path permits UI cancellation when no worker is
|
|
168
|
+
available; Windows timer overhead makes it slower in this test. Plain-text
|
|
169
|
+
streaming trades some throughput for lower memory even without those yields.
|
|
170
|
+
|
|
171
|
+
Reproduce with `node scripts/benchmark-profile.mjs <path> import`,
|
|
172
|
+
`text-stream-import` (worker-style scheduling), or `text-stream-main-import`
|
|
173
|
+
(cooperative scheduling). Do not compare these folded-stack figures with PLC
|
|
174
|
+
duration-trace timings as if they used the same input or parser.
|
|
@@ -0,0 +1,37 @@
|
|
|
1
|
+
# Java thread-dump snapshots
|
|
2
|
+
|
|
3
|
+
Import traditional HotSpot text output from `jstack PID` or
|
|
4
|
+
`jcmd PID Thread.print` by opening a `.jstack` file or a text file containing
|
|
5
|
+
the `Full thread dump …` header. Concatenate captures from the same JVM to
|
|
6
|
+
compare observed stacks across repeated snapshots. You can also open a ZIP of
|
|
7
|
+
traditional text captures from the same JVM; see [archive limits and supported
|
|
8
|
+
ZIP formats](JAVA_THREAD_DUMP_ZIP.md).
|
|
9
|
+
|
|
10
|
+
Each observed Java stack contributes one raw sample. The importer creates an
|
|
11
|
+
aggregate profile per thread name and Java thread ID, falling back to `tid`
|
|
12
|
+
when the Java ID is absent. Renamed threads appear separately. Missing threads
|
|
13
|
+
and threads without Java frames add no samples. The profile selector switches
|
|
14
|
+
between threads; Left Heavy, Sandwich and Call Tree show the observations.
|
|
15
|
+
Time Order is unavailable because snapshots do not measure execution duration.
|
|
16
|
+
CPU, elapsed, timestamps, state and lock text are not converted into weights.
|
|
17
|
+
|
|
18
|
+
Frames retain method/module names and available source files and positive line
|
|
19
|
+
numbers. Native Method and Unknown Source entries keep their identity without
|
|
20
|
+
inventing a source location. Deadlock summaries are ignored because they repeat
|
|
21
|
+
stacks already listed in the snapshot. Malformed thread headers, malformed Java
|
|
22
|
+
frames and duplicate thread identities within a snapshot produce an import error.
|
|
23
|
+
|
|
24
|
+
The parser iterates lines and retains the current stack and current snapshot's
|
|
25
|
+
thread identities, rather than collecting every line or observation. The text
|
|
26
|
+
input and resulting profile model still occupy memory; this format does not use
|
|
27
|
+
the duration-trace streaming importer. Prefer captures from one JVM per file.
|
|
28
|
+
The ZIP importer instead feeds bounded UTF-8 lines into the same parser and
|
|
29
|
+
checks both declared and actual decompressed sizes. Structured virtual-thread
|
|
30
|
+
JSON, native address stacks and lock/state visualization remain outside the
|
|
31
|
+
supported format. Unsupported ZIP variants can be extracted before importing
|
|
32
|
+
their text captures.
|
|
33
|
+
|
|
34
|
+
The checked-in [synthetic example](../sample/profiles/java-thread-dump/synthetic.jstack)
|
|
35
|
+
contains repeated snapshots and source locations. Format references are
|
|
36
|
+
[OpenJDK's traditional dump examples](https://github.com/openjdk/jdk/blob/master/test/hotspot/jtreg/serviceability/tmtools/jstack/utils/DefaultFormat.java)
|
|
37
|
+
and the [Oracle Java troubleshooting guide](https://docs.oracle.com/en/java/javase/21/troubleshoot/troubleshooting-guide.pdf).
|
|
@@ -0,0 +1,65 @@
|
|
|
1
|
+
# Java thread-dump ZIP archives
|
|
2
|
+
|
|
3
|
+
Open a ZIP containing traditional HotSpot `jstack` / `jcmd Thread.print` text
|
|
4
|
+
captures from **one JVM**. The same thread name and Java thread ID identify a
|
|
5
|
+
thread across snapshots; renamed threads remain separate. Each observed Java
|
|
6
|
+
stack contributes one raw sample, not measured elapsed time. Missing threads
|
|
7
|
+
and entries without Java stacks contribute no samples.
|
|
8
|
+
|
|
9
|
+
Entries are processed in deterministic, case-sensitive filename order, not ZIP
|
|
10
|
+
insertion order or file modification time. For example, `1.jstack`, `10.jstack`,
|
|
11
|
+
then `2.jstack` is the lexical order. The result is aggregate, so this ordering
|
|
12
|
+
does not claim a timeline or attach elapsed durations to snapshot intervals.
|
|
13
|
+
Every `.jstack` entry must contain its own `Full thread dump` header near the
|
|
14
|
+
start. Each entry boundary closes the previous snapshot.
|
|
15
|
+
|
|
16
|
+
The importer recognizes `.jstack`, `.txt`, `.log`, `.dump` and `.tdump` entries
|
|
17
|
+
with a `Full thread dump` header within the first 64 KiB. Other regular files
|
|
18
|
+
and ordinary notes are ignored as profile input, but still undergo bounded
|
|
19
|
+
decompression and checksum validation. Directories are skipped. Text is strict
|
|
20
|
+
UTF-8; UTF-16 dumps should be converted or imported as separate text files.
|
|
21
|
+
|
|
22
|
+
## Resource policy
|
|
23
|
+
|
|
24
|
+
| Resource | Maximum |
|
|
25
|
+
| --- | ---: |
|
|
26
|
+
| Compressed archive | 32 MiB |
|
|
27
|
+
| Archive entries, including directories | 512 |
|
|
28
|
+
| Decompressed bytes per entry | 16 MiB |
|
|
29
|
+
| Total decompressed bytes, including ignored files | 64 MiB |
|
|
30
|
+
| UTF-8 filename bytes | 512 |
|
|
31
|
+
| Path depth | 8 components |
|
|
32
|
+
| Decoded line length | 65,536 UTF-16 code units |
|
|
33
|
+
|
|
34
|
+
Declared sizes are checked before decompression. Actual expanded bytes are
|
|
35
|
+
checked on every output chunk before decoding or parsing, so lying ZIP headers
|
|
36
|
+
cannot bypass the limits. The inflater emits chunks of at most 16 KiB; a limit
|
|
37
|
+
violation can allocate that one output chunk before it is rejected. It never
|
|
38
|
+
accumulates the whole expanded entry. The line decoder retains at most one
|
|
39
|
+
bounded line, while the parser retains the current stack, current snapshot's
|
|
40
|
+
thread identities and the resulting profile builders. The compressed input and
|
|
41
|
+
final profile model still occupy memory; this is not an out-of-core viewer.
|
|
42
|
+
The importer yields during work and supports cancellation without returning a
|
|
43
|
+
partially imported group.
|
|
44
|
+
|
|
45
|
+
## Supported ZIP subset
|
|
46
|
+
|
|
47
|
+
Single-disk ZIP32 archives using **STORE** or **DEFLATE** are supported, including
|
|
48
|
+
ordinary data descriptors. Local/central headers, entry extents, decompressed
|
|
49
|
+
sizes and CRC32 checksums must agree. Paths are validated without extracting
|
|
50
|
+
anything to disk. Absolute/traversal paths, duplicate or case-colliding names,
|
|
51
|
+
file/directory conflicts, symlinks, special files and non-normalized Unicode
|
|
52
|
+
paths are rejected. Names must be ASCII or explicitly UTF-8.
|
|
53
|
+
|
|
54
|
+
Encrypted archives, ZIP64, split archives, self-extracting prefixes, archive
|
|
55
|
+
padding, alternate Unicode-path extra fields and other compression methods are
|
|
56
|
+
unsupported. Extract an unsupported archive yourself and open its text captures
|
|
57
|
+
individually or concatenate them into one `.jstack` file.
|
|
58
|
+
|
|
59
|
+
The checked-in [synthetic archive](../sample/profiles/java-thread-dump/synthetic.zip)
|
|
60
|
+
contains two short snapshots of the same thread. This is an independently
|
|
61
|
+
implemented ZIP32 reader based on
|
|
62
|
+
[PKWARE's ZIP specification](https://pkware.cachefly.net/webdocs/casestudies/APPNOTE.TXT),
|
|
63
|
+
using the existing
|
|
64
|
+
[pako incremental inflater](https://github.com/nodeca/pako/blob/1.0.6/lib/inflate.js).
|
|
65
|
+
It does not use JSZip's whole-entry `async('string')` allocation path.
|
|
@@ -1,5 +1,77 @@
|
|
|
1
1
|
# Large trace performance follow-up
|
|
2
2
|
|
|
3
|
+
## Long-timeline GPU precision (2026-11-21)
|
|
4
|
+
|
|
5
|
+
The renderer previously uploaded absolute time coordinates as float32. Near
|
|
6
|
+
6,000 seconds in a microsecond recording, endpoints 100 microseconds apart could
|
|
7
|
+
round to the same GPU value, making a call disappear at high zoom. A synthetic
|
|
8
|
+
[late-call recording](../sample/profiles/speedscope/long-late-calls.speedscope.json)
|
|
9
|
+
reproduces that failure, including early and late rectangles in one wide batch.
|
|
10
|
+
|
|
11
|
+
Rectangles are now clipped against the current atlas/source tile using CPU
|
|
12
|
+
doubles, then normalized before upload. The shader transforms unit coordinates
|
|
13
|
+
to the destination. Profile times, range-tree bounds, atlas keys, hit testing,
|
|
14
|
+
frame colors and measured durations retain their existing identities and values.
|
|
15
|
+
Each batch keeps one buffer and one full source-rectangle cache key; a shared
|
|
16
|
+
staging allocation is capped at 1,200,000 bytes for 10,000 rectangles. Switching
|
|
17
|
+
an uncached tile can require another bounded CPU normalization/upload; cached
|
|
18
|
+
tiles and unchanged source rectangles reuse their data. Final profile and GPU
|
|
19
|
+
buffer storage still scale with the recording.
|
|
20
|
+
|
|
21
|
+
Five regressions check real uploaded float32 coordinates and the renderer's
|
|
22
|
+
uniform transform, adjacent late calls, negative/spanning clipping, source
|
|
23
|
+
origin/extent changes, shared staging ownership, buffer lifecycle and limits.
|
|
24
|
+
Browser QA showed both 100-microsecond calls near 6,000 seconds, including after
|
|
25
|
+
overview reset/refocus. The private PLC imported four profiles and rendered
|
|
26
|
+
Time Order, Left Heavy and Sandwich's inverted callers/callees correctly during
|
|
27
|
+
pan/zoom checks, with no browser errors.
|
|
28
|
+
|
|
29
|
+
A warmed Node smoke using a preallocated-copy mock upload and source rectangles
|
|
30
|
+
covering almost all 10,000 rectangles measured about 0.19 ms per repack and
|
|
31
|
+
2.6 ms for 100,000 unchanged-source requests, with one buffer allocation. These are local CPU
|
|
32
|
+
measurements, not browser/GPU frame-rate claims. This addresses the demonstrated
|
|
33
|
+
float32 failure associated with [#467](https://github.com/jlfwong/speedscope/issues/467);
|
|
34
|
+
the original upstream capture remains unavailable for direct comparison. These
|
|
35
|
+
buffer/matrix tests also extend [#125](https://github.com/jlfwong/speedscope/issues/125),
|
|
36
|
+
but do not replace a browser/GPU pixel-regression harness.
|
|
37
|
+
|
|
38
|
+
Final local verification passed all 418 Jest tests in 72 suites with 307
|
|
39
|
+
snapshots on checksum-verified Node 22.23.5 ARM64 using a clean Jest cache.
|
|
40
|
+
The five Node tests, type checking, lint, schema and backlog checks passed
|
|
41
|
+
separately on the system Node runtime. Two full Jest runs on local Node 24.19.0
|
|
42
|
+
ARM64 exited with native access violations without an assertion failure, so a
|
|
43
|
+
green local Node 24 full-suite result is not claimed. GitHub's Windows/Linux
|
|
44
|
+
Node 22/24 matrix validates the branch independently before merge.
|
|
45
|
+
|
|
46
|
+
## Fixed-height Sandwich list geometry (2026-11-21)
|
|
47
|
+
|
|
48
|
+
Sandwich already rendered a viewport window. It now calculates that window and
|
|
49
|
+
initial scroll offset directly from row count/height, removing the size object
|
|
50
|
+
per frame and full-list geometry scans. A single cached selected-frame/index
|
|
51
|
+
replaces the frame-index Map: repeated, adjacent and wrap-around navigation are
|
|
52
|
+
constant time; unrelated new selections scan the existing sorted array once.
|
|
53
|
+
Missing results are cached. Element resizing updates the viewport and filtering
|
|
54
|
+
clamps stale scroll offsets, even without a browser-generated scroll event.
|
|
55
|
+
|
|
56
|
+
Tests cover 100,000-row selection/navigation, billion-row arithmetic without a
|
|
57
|
+
row allocation, visible-row bounds, filter shrink, resize and cleanup. The
|
|
58
|
+
billion-row case tests arithmetic only: browser scroll-height limits still apply.
|
|
59
|
+
The sorted row array, model, sorting and graph/color geometry remain proportional
|
|
60
|
+
to the recording. These changes reduce auxiliary storage and repeated work;
|
|
61
|
+
they do not establish a fully bounded-memory viewer.
|
|
62
|
+
|
|
63
|
+
Combined validation passed 410 Jest tests in 71 suites (307 snapshots), five
|
|
64
|
+
Node tests, type checking, lint, schema/backlog checks and both builds. The
|
|
65
|
+
private PLC gzip still exported four profiles, 17,510 frames and 1,039,656 call
|
|
66
|
+
events with the established SHA-256
|
|
67
|
+
`522de77afbfc4dd94f1f30c3a9551979d12db506305ffef999db9c910ae8eddb`.
|
|
68
|
+
The local Node streaming smoke took about 4.3 seconds to import; this is not a
|
|
69
|
+
browser performance guarantee or a claim of faster importing in this batch.
|
|
70
|
+
In browser QA the first PLC profile's 16,231-frame table rendered 40 rows at the
|
|
71
|
+
scroll end, clamped to zero when filtered to one row, and updated to 49 rendered
|
|
72
|
+
rows after a viewport resize. A generated local function link restored its
|
|
73
|
+
selection after reopening the same PLC recording in another tab.
|
|
74
|
+
|
|
3
75
|
Baseline: `eb93425`, reviewed 2026-11-18. These changes continue the open-issue
|
|
4
76
|
review, prioritizing large profiles and the export request in issue #89.
|
|
5
77
|
|
|
@@ -86,9 +158,261 @@ completed download, disabled Time Order for aggregates and combined-stack
|
|
|
86
158
|
inversion. The downloaded JSON was compared exactly with the standard exporter,
|
|
87
159
|
then reimported with all profile totals preserved.
|
|
88
160
|
|
|
89
|
-
|
|
90
|
-
|
|
91
|
-
|
|
92
|
-
|
|
93
|
-
|
|
94
|
-
and
|
|
161
|
+
## Worker import and search follow-up
|
|
162
|
+
|
|
163
|
+
Large input files (at least 1 MiB of input bytes) named `.json`, `.trace`,
|
|
164
|
+
`.cpuprofile` or `.heapprofile`, optionally gzip-compressed, now use a browser
|
|
165
|
+
worker when available. The existing format detector is reused, preserving its
|
|
166
|
+
dispatch order. HTTP/profileURL, local File and CLI base64 loading use this path.
|
|
167
|
+
Input buffers and numeric profile tables transfer ownership instead of being
|
|
168
|
+
JSON-serialized or cloned as full call-tree objects. File inputs can be read in
|
|
169
|
+
the worker. Symbols are demangled there before returning immutable profile data.
|
|
170
|
+
|
|
171
|
+
The UI shows import progress and cancellation. Cancellation terminates the worker;
|
|
172
|
+
new loads and unmounts cancel old imports, and stale results cannot replace newer
|
|
173
|
+
profiles or clear their loading state. The worker is released before UI object
|
|
174
|
+
reconstruction, which yields between bounded batches so it can also be cancelled.
|
|
175
|
+
Worker creation blocked by a host can fall back to the ordinary importer. An
|
|
176
|
+
unexpected worker crash reports an error rather than trying a detached buffer.
|
|
177
|
+
Small files and other filename extensions retain the ordinary importer; that
|
|
178
|
+
synchronous fallback cannot be interrupted during parsing.
|
|
179
|
+
|
|
180
|
+
The worker code is embedded by the build plugin, so both HTTP and offline builds
|
|
181
|
+
can create it without fetching a worker script or WASM from an external host.
|
|
182
|
+
The transport is an internal implementation detail, not a public interchange API.
|
|
183
|
+
|
|
184
|
+
Tests cover numeric/string frame identity, recursion, source locations, formatting,
|
|
185
|
+
aggregate/derived flags, idle ranges, transferred buffers, deep trees, invalid
|
|
186
|
+
references, cancellation, worker cleanup, host fallback and stale-load races.
|
|
187
|
+
The PLC worker import was cancelled interactively while parsing, then completed;
|
|
188
|
+
its exported JSON matched the ordinary importer exactly (four profiles, 17,510
|
|
189
|
+
frames and 1,039,656 call events). A local snapshot smoke check measured 214 ms to
|
|
190
|
+
encode the PLC profiles and 34,602,732 bytes of numeric tables (excluding frame
|
|
191
|
+
strings and metadata). Browser elapsed-time/peak-RSS speedups are not claimed.
|
|
192
|
+
|
|
193
|
+
Search also supports literal whole-word and exact matching. Patterns are reused
|
|
194
|
+
per query and file-only checks avoid allocating name highlight ranges. Unicode
|
|
195
|
+
word boundaries, punctuation, escaped metacharacters and full-name/path matching
|
|
196
|
+
have regressions. User-provided regex is still not enabled.
|
|
197
|
+
|
|
198
|
+
Profile trees and rendering remain in-memory operations. Unsupported formats
|
|
199
|
+
still use the original full-document importer. UI hydration and rendering have
|
|
200
|
+
costs, and transferring work can increase overall elapsed time even while making
|
|
201
|
+
the interface responsive. This does not fix every OOM report. Derived
|
|
202
|
+
transformations other than frame filtering remain synchronous. Frame filtering
|
|
203
|
+
visits grouped paths and yields during traversal, with cancellation; final tree
|
|
204
|
+
sorting remains synchronous. See [filter behavior and limits](PROFILE_FILTERS.md).
|
|
205
|
+
Export retains the final
|
|
206
|
+
Blob bytes and source profile while bounding temporary serialization chunks
|
|
207
|
+
(apart from an individually huge string). An out-of-core model, heap budgets,
|
|
208
|
+
worker-backed transforms, signed comparisons and synchronized timelines remain
|
|
209
|
+
separate follow-ups in [the maintained backlog](ISSUE_BACKLOG.md).
|
|
210
|
+
|
|
211
|
+
## Streaming duration traces
|
|
212
|
+
|
|
213
|
+
Large duration-event `.json`/`.trace` files, optionally gzip-compressed, now use
|
|
214
|
+
incremental parsing inside the worker. A local File is streamed without reading
|
|
215
|
+
its whole contents; transferred ArrayBuffers are consumed in 64 KiB slices.
|
|
216
|
+
Native `DecompressionStream` inflates gzip incrementally. The streaming parser
|
|
217
|
+
([@streamparser/json](https://github.com/juanjoDiaz/streamparser-json), pinned at
|
|
218
|
+
0.0.26) releases completed event objects from its parent array. We retain the
|
|
219
|
+
duration events, naming metadata and first event needed by the existing Hermes
|
|
220
|
+
discriminator, dropping unused event fields and ignored events. Event arguments
|
|
221
|
+
remain intact because they participate in frame identities.
|
|
222
|
+
|
|
223
|
+
This avoids a full inflated byte buffer, a whole JSON string and a complete
|
|
224
|
+
parsed input object. Out-of-order traces still need stable sorting, so retained
|
|
225
|
+
events and constructed trees continue to scale with the capture. This is not an
|
|
226
|
+
out-of-core profile store or a hard bound on total JavaScript/browser memory.
|
|
227
|
+
|
|
228
|
+
The streaming path rejects excessive input complexity before returning a profile:
|
|
229
|
+
|
|
230
|
+
| Limit | Default |
|
|
231
|
+
| --- | ---: |
|
|
232
|
+
| Decompressed input | 1 GiB |
|
|
233
|
+
| Pending event/metadata bytes between completed events | 16 MiB |
|
|
234
|
+
| Retained event source bytes | 256 MiB |
|
|
235
|
+
| Input events, including ignored events | 2,000,000 |
|
|
236
|
+
| JSON nesting depth | 128 |
|
|
237
|
+
|
|
238
|
+
Checks run before bounded parser writes and while tokens/events are emitted.
|
|
239
|
+
The pending-byte check is conservative at chunk boundaries. These are source
|
|
240
|
+
complexity limits, not heap accounting: object overhead, expanded Unicode,
|
|
241
|
+
sorting, call-tree nodes, numeric snapshots, hydration and GPU resources have
|
|
242
|
+
additional costs. A limit failure stops with an actionable error; it never
|
|
243
|
+
silently truncates the trace or retries the full importer. Cancellation releases
|
|
244
|
+
the reader and terminates the worker; source cancellation failures cannot hide
|
|
245
|
+
the original error.
|
|
246
|
+
|
|
247
|
+
Native CPU/heap/speedscope files and known filename overrides retain their
|
|
248
|
+
detector. CPU markers anywhere in the stream, sample tables, duplicate root
|
|
249
|
+
`traceEvents` properties and unrecognized wrapper properties return to that
|
|
250
|
+
detector, preserving their original semantics. Those fallback formats do not
|
|
251
|
+
gain these streaming budgets. Gzip also falls back when native decompression is
|
|
252
|
+
unavailable. URL/CLI loading still buffers the input response before transfer;
|
|
253
|
+
local File streaming avoids that buffer. Tolerated trailing commas and missing
|
|
254
|
+
final `]` after complete array events remain supported. Incomplete values and
|
|
255
|
+
corrupt gzip fail rather than producing a partial result.
|
|
256
|
+
|
|
257
|
+
Validation includes every duration-trace fixture against the ordinary importer,
|
|
258
|
+
one-byte UTF-8/escape boundaries, late metadata, argument identity, unordered and
|
|
259
|
+
anonymous events, sampled/CPU fallback, ignored payloads, gzip File imports with
|
|
260
|
+
whole-file reads forbidden, resource-limit failures, and cancellation races.
|
|
261
|
+
The PLC capture exported identically: four profiles, 17,510 frames, 1,039,656
|
|
262
|
+
call events and matching totals. Fresh-process Windows/Node 24 import-only runs:
|
|
263
|
+
|
|
264
|
+
| Path | Import time | Peak RSS | Heap after import |
|
|
265
|
+
| --- | ---: | ---: | ---: |
|
|
266
|
+
| Ordinary importer | 2,281 ms | 999 MiB | 572 MiB |
|
|
267
|
+
| Streaming duration importer | 4,311 ms | 623 MiB | 437 MiB |
|
|
268
|
+
|
|
269
|
+
This single capture/run measured about 38% lower peak RSS at a latency cost.
|
|
270
|
+
Browser peak memory, rendering and worker hydration are not represented by this
|
|
271
|
+
Node benchmark. Results vary by machine, input and garbage collection. Reproduce
|
|
272
|
+
each path in its own process with `node scripts/benchmark-profile.mjs <capture>
|
|
273
|
+
import` and `node scripts/benchmark-profile.mjs <capture> stream-import`.
|
|
274
|
+
|
|
275
|
+
Issues #519, #347, #478, #386 and #404 remain **partial** in the backlog: the
|
|
276
|
+
specific original reproductions and a general bounded-memory profile model still
|
|
277
|
+
need work.
|
|
278
|
+
|
|
279
|
+
## Faster streaming event parsing
|
|
280
|
+
|
|
281
|
+
CPU profiling of the first streaming implementation showed substantial time in
|
|
282
|
+
token parsing, per-property assignment, selector matching and callbacks. The
|
|
283
|
+
worker now scans structural bytes to frame each complete event object, then
|
|
284
|
+
decodes and parses that one object with native `JSON.parse`. The token parser
|
|
285
|
+
still validates the wrapper/array grammar and handles other values. The bounded
|
|
286
|
+
event framing scan tracks string escapes and nesting; it does not interpret
|
|
287
|
+
fields or split on text delimiters. A rare permissive retry retains the existing
|
|
288
|
+
trailing-comma behavior.
|
|
289
|
+
|
|
290
|
+
No whole-document string or inflated buffer is introduced. Events spanning read
|
|
291
|
+
boundaries use a temporary buffer for that one event. The existing decompressed,
|
|
292
|
+
pending, retained, event-count and depth limits remain in force, including while
|
|
293
|
+
framing native records. Truncated later records cannot be silently dropped when
|
|
294
|
+
repairing a producer's omitted final array bracket. Additional regressions cover
|
|
295
|
+
all small chunk sizes, large records, split UTF-8/escapes, malformed inner and
|
|
296
|
+
outer grammar, and the existing fallback/cancellation/limit cases.
|
|
297
|
+
|
|
298
|
+
Three alternating fresh-process runs of the same PLC capture on Windows/Node 24
|
|
299
|
+
compared the previous streaming bundle (`39b264b`) with native event framing:
|
|
300
|
+
|
|
301
|
+
| Metric | Previous streaming | Native event framing |
|
|
302
|
+
| --- | ---: | ---: |
|
|
303
|
+
| Import times | 4,255 / 4,336 / 4,423 ms | 2,493 / 2,563 / 2,445 ms |
|
|
304
|
+
| Median import time | 4,336 ms | 2,493 ms |
|
|
305
|
+
| Peak RSS | 612 / 613 / 615 MiB | 579 / 573 / 581 MiB |
|
|
306
|
+
| Median peak RSS | 613 MiB | 579 MiB |
|
|
307
|
+
|
|
308
|
+
That is about 43% faster while retaining (and slightly improving) the previous
|
|
309
|
+
memory reduction. These are import-only process measurements, not browser
|
|
310
|
+
rendering or hydration measurements. Each revision's `benchmark-profile.ts` was
|
|
311
|
+
bundled separately with esbuild (Node/CommonJS, ES2020, binary WASM) and run with
|
|
312
|
+
the capture path and `stream-import` action, alternating revisions without
|
|
313
|
+
explicit garbage collection. Captures and timing logs are not committed.
|
|
314
|
+
The full exported profiles match the ordinary importer, including all four
|
|
315
|
+
profiles, 17,510 frames, 1,039,656 call events and profile totals.
|
|
316
|
+
|
|
317
|
+
The same follow-up fixes issue #199's deep-zoom axis labels: formatting receives
|
|
318
|
+
the visible tick interval, tick generation clamps below floating-point precision,
|
|
319
|
+
and overlapping labels are skipped. V8 source-path extraction and exact-first
|
|
320
|
+
sourcemap matching (with empty maps ignored) improve #446, which remains partial
|
|
321
|
+
because automatic linked map discovery/loading is still separate work. The
|
|
322
|
+
maintained backlog records these completion boundaries explicitly.
|
|
323
|
+
|
|
324
|
+
## Smaller duration queues and bounded selected-stack details
|
|
325
|
+
|
|
326
|
+
The 2026-11-20 follow-up reuses complete (`X`) events as begin records rather
|
|
327
|
+
than spreading every event into another full object. Synthesized end records
|
|
328
|
+
retain only their timestamp, name and shared arguments. Source positions and
|
|
329
|
+
argument-based frame identity remain unchanged. Queues already ordered by time
|
|
330
|
+
skip stable-sort workspace; unordered queues still use the existing stable sort,
|
|
331
|
+
including equal-timestamp and epoch-rounding behavior. No additional event index
|
|
332
|
+
is retained. The streaming limits are unchanged.
|
|
333
|
+
|
|
334
|
+
Three alternating fresh-process runs on Windows/Node 24 compared `f6c5d43`
|
|
335
|
+
against this change, using the same ES2020 CommonJS benchmark bundle options:
|
|
336
|
+
|
|
337
|
+
| Capture / metric | Before | After |
|
|
338
|
+
| --- | ---: | ---: |
|
|
339
|
+
| PLC import times (ms) | 2,609 / 2,524 / 2,579 | 2,562 / 2,546 / 2,562 |
|
|
340
|
+
| PLC median peak RSS (MiB) | 576 | 576 |
|
|
341
|
+
| Complete-event fixture import times (ms) | 1,131 / 833 / 822 | 986 / 767 / 779 |
|
|
342
|
+
| Complete-event fixture median peak RSS (MiB) | 362 | 295 |
|
|
343
|
+
| Complete-event fixture median heap at import (MiB) | 224 | 155 |
|
|
344
|
+
|
|
345
|
+
The synthetic fixture contains 250,000 sequential complete events with 5,000
|
|
346
|
+
distinct argument-based frame keys. Its median peak RSS fell about 19%; PLC
|
|
347
|
+
performance remained essentially unchanged. These are process import measurements,
|
|
348
|
+
not browser hydration/rendering or hard heap bounds. Separate full-export checks
|
|
349
|
+
matched the baseline hash for both captures. PLC still exports four profiles,
|
|
350
|
+
17,510 frames and 1,039,656 call events. Captures and benchmark artifacts are not
|
|
351
|
+
committed. Reproduce import measurements with `node scripts/benchmark-profile.mjs
|
|
352
|
+
<capture> stream-import`; do not include export validation in import-only RSS.
|
|
353
|
+
|
|
354
|
+
Selected-stack details now retain/render at most 100 ancestor rows and follow
|
|
355
|
+
parent pointers only as pages are requested, without building the full ancestor
|
|
356
|
+
list. All scopes remain reachable through older-page and reset controls. Clicking
|
|
357
|
+
one selects the exact ancestor instance; see [stack navigation](SEARCH_AND_LAYOUT.md#navigate-selected-stack-ancestors).
|
|
358
|
+
Tests include a 100,000-level chain with a guarded page boundary. The PLC's
|
|
359
|
+
deepest seven-scope stack was paged without changing the export hash.
|
|
360
|
+
|
|
361
|
+
Issues #519, #478, #386, #347 and #404 remain partial: the final profile model,
|
|
362
|
+
worker snapshots, unordered sorts and graph rendering still grow with capture
|
|
363
|
+
size. This work reduces specific temporary allocations and bounds the details
|
|
364
|
+
panel, without claiming an out-of-core viewer or fixing every reported OOM.
|
|
365
|
+
|
|
366
|
+
Validation passed 339 Jest tests across 58 suites (307 snapshots), five Node
|
|
367
|
+
tests, type checking, linting, schema/backlog checks, both builds and a dry-run
|
|
368
|
+
npm contents check. Browser smoke checks used a 220-level profile to verify
|
|
369
|
+
100-row pages, access to the root, reset controls, Enter/Space selection and
|
|
370
|
+
visible ancestor highlighting below the time axis, without console errors.
|
|
371
|
+
|
|
372
|
+
## Bounded graph-search state
|
|
373
|
+
|
|
374
|
+
On 2026-11-21, graph search stopped eagerly caching every source symbol and
|
|
375
|
+
allocating navigation geometry/index entries for every matching call. Per query,
|
|
376
|
+
boolean frame results are capped at 4,096 entries and drawn-label ranges at
|
|
377
|
+
512 entries with at most 64 ranges each. Time Order/Left Heavy navigation keeps
|
|
378
|
+
the first 10,000 matches and checks for one further match to establish
|
|
379
|
+
truncation. The search UI displays a `+` lower bound and a visible refinement
|
|
380
|
+
notice; it never claims the truncated count is complete. Matching/highlighting
|
|
381
|
+
can continue beyond that navigation window.
|
|
382
|
+
|
|
383
|
+
Three alternating fresh-process Windows/Node 24 runs compared `a72d309` with
|
|
384
|
+
bounded search on the PLC's first profile (513,009 graph frames), using a broad
|
|
385
|
+
single-space query:
|
|
386
|
+
|
|
387
|
+
| Search-only metric | Before | After |
|
|
388
|
+
| --- | ---: | ---: |
|
|
389
|
+
| Initialization times (ms) | 96 / 89 / 94 | 3 / 3 / 4 |
|
|
390
|
+
| Retained search heap after explicit GC (MiB) | 119.60 / 119.60 / 119.60 | 0.907 / 0.907 / 0.906 |
|
|
391
|
+
| Retained navigation matches | 513,009 | 10,000 (explicitly limited) |
|
|
392
|
+
|
|
393
|
+
This is a deliberate navigation-limit tradeoff, not equivalent full-result
|
|
394
|
+
enumeration. Separate single-run checks found the same 40 `GroupCalls` matches
|
|
395
|
+
(14 → 22 ms) and zero matches for a nonexistent symbol (14 → 25 ms). Those sparse
|
|
396
|
+
queries retain about half as much search heap, but can be slower because cached
|
|
397
|
+
entries are recomputed. These are Node search measurements after import and graph
|
|
398
|
+
construction, not whole-app memory caps or browser latency guarantees.
|
|
399
|
+
The measured initialization counts results and opens the first ten; it does not
|
|
400
|
+
draw labels or measure browser rendering.
|
|
401
|
+
|
|
402
|
+
The PLC import/export check still reports four profiles, 17,510 source frames
|
|
403
|
+
and 1,039,656 call events, with the unchanged baseline export hash. Captures and
|
|
404
|
+
benchmark artifacts remain private/uncommitted. The maintained benchmark adds
|
|
405
|
+
`node scripts/benchmark-profile.mjs <capture> stream-search [query]` for streaming
|
|
406
|
+
import followed by bounded search of the first profile; its default query is a
|
|
407
|
+
single space. It reports the navigation limit and retained search heap separately
|
|
408
|
+
from input/model memory. Graph creation, worker snapshots, sorting, Sandwich
|
|
409
|
+
row lists and the final model remain proportional to recording size; the OOM
|
|
410
|
+
backlog stays partial.
|
|
411
|
+
|
|
412
|
+
Validation passed all 354 Jest tests across 62 suites (307 snapshots), five Node
|
|
413
|
+
tests, type checking, linting, schema/backlog checks, both builds and package
|
|
414
|
+
contents. Regressions cover cache eviction, dense-symbol range limits, exact
|
|
415
|
+
limit detection, late sparse matches, depth-first order and 25,000-level graphs.
|
|
416
|
+
Browser checks on 20,050 calls verified the visible limit notice, both keyboard
|
|
417
|
+
wrap directions, narrowing to a late result, and the root-label axis fix,
|
|
418
|
+
without console errors.
|