[ccoventry]

Release note: Fixed: the demo loading progress in the game fills all the way on a demo that recorded a map restart, instead of disappearing at about half.

Closes #866

Cause

The demo player's clock is svc_time, and viewdemo stops loading where that clock turns back at a map restart (it only plays a demo's first map). The progress divided the player's span by the directory's whole playback time, less the LOADING lead-in (#810). On solifotw-bewton the player holds 4.30 → 1956.38 s, against 2,996 s in the directory, so the bar went away at about 53%. Measured with an R&D build; the numbers are on #866.

Change

• demo_file::loaded_span: walks every frame header and returns the svc_time span from the first network frame to the first backwards jump. It reads the file through a 1 MB BufReader with seek_relative, so the buffer survives the skips.
• load_progress: runs it on a background thread when a demo starts loading. load_percent then divides the player's span by it, the same clock with no lead-in. Until the thread answers, the directory's estimate stands.
• The per-frame path only uses try_lock.

Tests

• New unit tests: the_loaded_span_ends_where_the_server_clock_turns_back (a built demo file with a restart) and with_the_loaded_span_the_percent_is_the_players_span_over_it. cargo test -p goldsrc-hooks --target i686-pc-windows-msvc: 678 passed.

• Clippy 1.98.1 -D warnings is clean, both i686 and host. fmt is clean.

• In game (game_probe, Brain account, DLL local\test-866\, the Playback tab's 10% log lines):

  • POST, solifotw-bewton (the restart demo): 18, 20, 30 … 90%, then loaded. It used to stop at 50%.

  • POST, solifotw-bud (no restart): 11, 20 … 90%, then loaded.

  • PRE, wsod25_grp2_h1_hltv: 10 … 90%, then loaded.

  • Re-run after merging dev (In-game commands get shorter names grouped by what they change: ds_hud_hide_scoreboard and so on (#555) #719), solifotw-bewton on PRE and POST, DLL rebuilt into local\test-866\: 0, 17/18, 20 … 90%, then loaded on both. Reports: local\game-probe\875-ds-pre\, 875-ds-post\. After the merge: hook tests (689), workspace tests, clippy (both targets), fmt and npm run test:unit pass.

  On the restart demo the first line is 18%: the directory's estimate showed until the file walk answered, about 3 s into the load.

🤖 Generated with Claude Code

[ccoventry]

Merged dev in (after #719) at 4563a4b5, with no conflicts. This PR adds and uses no console names.

• Local checks: fmt, clippy 1.98.1 (workspace and i686), hook tests (689), workspace tests, npm run test:unit: all pass.
• Test DLL rebuilt: local\test-866\dodstudio_goldsrc_hooks.dll.
• game_probe re-run with viewdemo solifotw-bewton (the map-restart demo) on PRE and POST. The Playback tab logged 0, 17% (POST 18%), then 20 … 90%, then "the demo is loaded", on both. Both PASS (local\game-probe\875-ds-pre\, 875-ds-post\).

🤖 Generated with Claude Code

[ccoventry]

Brought up to date with dev (unattended pass). Head dd17cda1.

• Two gh pr update-branch runs, both clean.
• CI green on dd17cda1.

In game. DLL local\test-866\, rebuilt from dd17cda1. game_probe, Brain account, viewdemo solifotw-bewton (the map-restart demo).

• PRE, PASS (local\game-probe\uh-875-pre\ on f863886, uh-875-pre-r2\ on dd17cda):
  • Playback tab logged 0, 18, 20, 30 ... 90%, then "the demo is loaded".
  • ds_debug_status: temp-entity fix 6 of 6, hull-trace guard on (pre-Anniversary on PRE, 25th Anniversary on POST)
• POST, PASS (uh-875-post\, uh-875-post-r2\): the same, 0, 18, 20 ... 90%, then loaded.
• Neither hook log shows a hook failing. The only "not installed" lines are modules that are build-specific by design: frame_esc on PRE, lightmap_gamma on POST.

🤖 Generated with Claude Code