The live TUI now knows the terminal size and says more while a scan runs.

• the feed and the node panel are sized to the terminal window, and no longer
  push the header and the finished phases off screen
• f gives the feed the whole window, n expands the node panel past its cap
• the progress bar carries working/torn/node counts and elapsed time; a finished
  phase says how long it took
• the feed and the node panel go once the run is done, the console tables and
  the report already repeat them
• plain (non-TTY) output prints scan progress instead of going quiet for minutes

Commits

• feat: print scan progress in plain output mode (deb82ac)
• fix: put the elapsed time after the bar counters (a27ba70)
• perf: insert into the live feed instead of resorting it (1ea0796)
• feat: show elapsed time on the bar and the phase lines (0d9503e)
• feat: show working, torn and node counts next to the bar (3560411)
• fix: drop the node panel with the feed, the report prints Nodes too (efb319b)
• fix: drop the live feed once the console tables repeat it (09b8bb2)
• fix: keep the feed at 12 rows unless f expands it (ecb1904)
• feat: add f and n to expand the live feed and the node panel (7be820e)
• feat: size the live scan feed to the terminal window (68d2035)
• feat: show every region and node found so far during the scan (9cb078b)

v0.14.0 Latest

Latest

Compare

Choose a tag to compare

• ENDPOINT PING was wrong on any host without unprivileged ICMP sockets:
  the raw socket sees every reply on the host, and the matcher only checked
  the sequence number, so a parallel scan mixed endpoints' replies together.
  Measured on OpenWrt: a Frankfurt endpoint 41ms away reported 6-32ms and a
  2ms one reported 0.09ms. Now the source address has to match too.
• the latest-version check caches its answer in the temp dir instead of
  ~/.cache, which on OpenWrt is the flash overlay.

Commits

• fix: drop host ping replies that came from another endpoint (8ec2902)
• fix: cache the latest-version answer in the temp dir, not on flash (073baf7)

v0.13.4

Compare

Choose a tag to compare

Six bug fixes, four of them reported and two of them fixed by @v14d4n.

• scan -conf FILE now exits non-zero when the config cannot be written,
  instead of printing the failure and exiting 0 (#6)
• -best and -conf - are rejected as a pair: both write a machine-readable
  artifact to stdout, and together they produced a config no client can
  import (#7)
• -n 0 and -n -1 are a usage error instead of silently scanning every
  address of every pool (#8)
• register no longer treats an unreadable or malformed account file as an
  absent one, so a rotation cannot turn into a fresh registration that
  overwrites it - -fresh still replaces it deliberately (#5)
• the per-subnet table sorts subnets with no endpoint below working ones,
  where a synthetic 0% loss used to float them to the top (#9)
• a failed masque-h2 scan says the network blocks MASQUE instead of
  suggesting -gen-i1, which does not apply to it (#10)

Commits

• fix: return the no-endpoint failure from printBest instead of exiting inside it (3eea39c)
• fix: stop register from overwriting an account file it could not read (closes #5) (b2903f2)
• fix: reject -best together with -conf - instead of mixing them on stdout (closes #7) (1daaacc)
• fix: sort the picks table by status before loss (closes #9) (9f6ebae)
• fix: tell a failed masque-h2 scan the same thing as masque (closes #10) (85f193a)
• fix: reject a nonpositive -sample instead of widening the scan (a78d29a)
• fix: fail the scan when a requested -conf cannot be written (430a8a1)
• ci: refetch the annotated tag before reading its body (6be5efb)

New Contributors

• @v14d4n made their first contribution in #11

Full Changelog: v0.13.2...v0.13.3

v0.13.3

Compare

Choose a tag to compare

Fixed a full scan being OOM-killed on small hosts, and cut what the
scan spends on CPU and on waiting.

• the Go runtime is capped at 32MiB, so scan -f no longer gets
  OOM-killed on a ~1 GB VPS
• every wg/awg tunnel now uses the batch-1 bind: same results, 11x
  less CPU on a full scan (304s -> 28s)
• phase 1 probes addresses in parallel - on a network where nothing
  answers, its worst case drops from ~22 minutes to 3m39s
• the host ping burst goes out at once: an endpoint that ignores ICMP
  costs one -timeout instead of three
• the handshake poll backs off from 2ms instead of ticking at a flat
  50ms, so a live endpoint is confirmed in ~7ms instead of ~51ms

Commits

• perf: back off the handshake poll instead of a flat 50ms tick (a346aa5)
• perf: resolve the meta address without a global lock (6557ecf)
• perf: send the host ping burst at once and match replies by sequence (b414d42)
• perf: probe phase 1 addresses in parallel instead of one at a time (9e1c446)
• perf: give every wg tunnel the batch-1 bind, not conn.NewDefaultBind (58c7551)
• fix: cap the Go runtime at 32MiB so a full scan is not OOM-killed (6668e26)
• ci: lead the release notes with the annotated tag body (b71edc5)
• ci: leave docs commits out of the release notes (899590c)

v0.13.2

Compare

Choose a tag to compare

Android has no /etc/resolv.conf, the only file Go's own resolver reads, so
every lookup failed with connection refused and register had nothing left
to try - not the direct API call, not the relay, not the tunnel fallback.
warpscout now reads Termux's own resolv.conf, falling back to 8.8.8.8.

Commits

• fix: resolve DNS on Android, where there is no /etc/resolv.conf (77e70f4)

v0.13.1

Compare

Choose a tag to compare

Android is a release target now, and every command tells you when a new
release is out.

• android/arm64 in the release matrix, cross-built with no NDK; install.sh
  detects Termux and installs into $PREFIX/bin
• every command but version checks GitHub for a newer release first, cached
  for 6h
• register grew a relay fallback (direct -> relay -> WARP tunnel), so it
  works on a network that blocks the Cloudflare API
• -port N pins one port for the whole scan - the only way to compare ports,
  since a normal run mixes them across endpoints
• the socks exit row no longer says the country twice

Commits

• refactor: keep the version cache as a single file in the cache dir (620fce4)
• feat: check for a new release on startup (c882667)
• docs: mention the registration relay in both READMEs (a203ca1)
• feat: add a relay fallback to register (e39e5a4)
• feat: print the macOS quarantine hint in install.sh (220dd06)
• docs: mention the Android build in both READMEs (d2f5390)
• feat: detect Termux in install.sh and install into $PREFIX/bin (f68b50e)
• ci: build android/arm64 in the release matrix (61e6a32)
• fix: strip the path termux-exec injects into argv on Android (558e6f8)
• docs: mention censorcheck alongside ipregion in the socks section (84e03ef)
• docs: cover masque as a third protocol and refresh the account file layout (ece832a)
• fix: stop repeating the country in the socks exit row (52dc225)
• docs: reflow the markdown tables in both READMEs (9569909)
• docs: document -port in both READMEs (0684de2)
• feat: add -port to pin one port for the whole scan (8930cbc)
• ci: skip the test run for workflow-only changes (0b8e3c3)
• ci: skip the test run for the release and docker workflows (4997d3f)
• ci: keep only the commit list and New Contributors in release notes (dc88f38)

v0.13.0

Compare

Choose a tag to compare

Commits

• docs: document the exclude filters in both READMEs (4cb718a)
• feat: add -exclude-node and -exclude-country (f0b0878)
• feat: give the console table a row per node, not just per subnet (8fcd9bd)
• feat: name the chained mihomo proxies after their place in the chain (3c86a5f)
• docs: document the two-interface -through config in both READMEs (916c51b)
• feat: write both tunnels of the chain into the native -through config (adf4998)
• docs: document the chained mihomo config in both READMEs (6df3ba0)
• feat: render the whole chain in the mihomo config under -through (eb02781)
• docs: document socks in both READMEs (0621fe0)
• feat: add socks to serve one endpoint as a local SOCKS5 proxy (a142752)
• docs: document scan -through in both READMEs (50251ce)
• feat: add scan -through to nest one WARP tunnel inside another (290aa1f)
• feat: carry both address families inside the WireGuard tunnel (e2e0361)
• feat: keep the WARP device's own IPv4 and IPv6 from registration (bdce03d)
• docs: document the wider MASQUE v6 pools in both READMEs (57adc0c)
• feat: add the second MASQUE v6 block, 103:: and 104:: (thanks @t-e-s-tweb) (6d3b891)
• docs: drop -jt, justify 8MiB (4b63d49)
• docs: document GOMEMLIMIT for OOM kills in both READMEs (9381c18)

New Contributors

• @labi-le made their first contribution in #4

Full Changelog: v0.11.1...v0.12.0

v0.12.0

Compare

Choose a tag to compare

Commits

• ci: bump the Docker build image to go 1.26 (dcb7706)

Full Changelog: v0.11.0...v0.11.1

v0.11.1

Compare

Choose a tag to compare

Commits

• ci: skip the test workflow on doc-only changes (e102a95)
• docs: document find-sni -p in both READMEs (12e3bbe)
• feat: let find-sni search an SNI for either MASQUE transport (d30063d)
• refactor: name the mihomo proxy after the protocol instead of the endpoint (634115e)
• docs: document masque-h2 in both READMEs (83c9410)
• feat: add -p masque-h2, CONNECT-IP over TCP with real endpoint pools (closes #3) (4281051)

Full Changelog: v0.10.0...v0.11.0

v0.11.0

Compare

Choose a tag to compare

Commits

• docs: document -speed in both READMEs (f53a3af)
• feat: add -speed to measure endpoint throughput after the scan (2a03132)
• refactor: extract nodePicks and the feed capping helpers for reuse (8dafee7)
• refactor: hoist the bubbletea program out of runScanUI into runWithUI (7303a35)
• docs: document find-sni in both READMEs (0dc8c3f)
• feat: add find-sni to search a MASQUE SNI that gets through (5c7aa92)
• perf: bound the stalled MASQUE dial with quic-go's idle timeout (3ac583c)
• docs: add a release downloads badge to both READMEs (4bc8231)
• docs: document install.sh and the version command in both READMEs (42e3abe)
• feat: add install.sh for Linux, macOS and OpenWrt (5e30e8d)
• feat: add a version command, with the tag injected at build time (35c4ea2)

Full Changelog: v0.9.0...v0.10.0

v0.10.0

Compare

Choose a tag to compare