Documentation of Race the Clock (Tandy, 1992) for the Tandy / Memorex Video
Information System, measured from a .bin/.cue pair. No disc image and no
product bytes are in this repository — only measurements, the tools that made
them, and a hash list.
| Title | Race the Clock, USA, English and Spanish |
| Year | 1992 — the mastering year, from the disc's own four internal dates, stated as such |
| Publisher | Tandy, from the ISO 9660 publisher identifier |
| Developer | Mindplay, Tucson, Arizona in CONTROL.TAT; Methods & Solutions, Inc., Tucson, Arizona in the program's own title-screen strings. Three fields, two companies |
| Image | 433,132,560 B, sha1 95f950fe625d0e375f325d033673ef876f75555c, one MODE1/2352 track, 184,155 sectors |
| Contents | 241 directories, 3,625 files, 374,094,820 B, six extensions |
| Distinct content | 1,314 distinct SHA-1, 73,748,704 B — 19.7139 %. The other 80.2861 % is copies |
| Video | 1,208 AVI, 95.8047 % of declared bytes; BI_RLE8 8-bit; two streams on 1,208 of 1,208; 22,050 Hz PCM throughout; 1 h 24 m 38 s of playable files over 13 m 12.53 s of distinct footage |
| Program | RTC.EXE, 136,192 B = 0.0364 %, a Win16 NE, two segments, 75.8121 % resources, imports KERNEL GDI USER WIN87EM MMSYSTEM TVUI HC |
| Is it a game? | Four of four on a criterion with a test attached — chapter 11 |
A bilingual vocabulary game. Forty rounds; each draws a distinct fifteen from a pool of sixty filmed action clips, and hides a hand-drawn 480 × 300 picture behind thirty 80 × 60 tiles. Nine levels, one player or two, and an opponent that is a clock. Every clause of that sentence has a byte behind it in chapter 11; none of it comes from a filename.
The disc is the second VIS pressing anybody has documented, thirteen days
after the first, from the same Tandy Maketat build, and it contradicts it in
the most useful direction: the first was a real-mode DOS binary and this one is
the Win16 NE application the platform notes always said a VIS title would be.
| 00 | 22 inherited clauses and 71 open ones, written before the image was opened |
| 01 | what the object is, its four denominators, and its coverage |
| 02 | every figure with the command that makes it again |
| 03 | Mode 1, five sectors that are not, and a mastering record that points at itself |
| 04 | the NE: seven modules, six window procedures, fifty-nine strings, forty-five padded paths |
| 05 | 1,208 Video for Windows files on a 1992 console, decoded |
| 06 | 2,405 DIB: forty drawings cut into thirty pieces each, reassembled |
| 07 | 80.2861 %, keyed on hashes and on names, both published |
| 08 | 765 bytes of .INI describing a console that is not on the disc |
| 09 | CONTROL.TAT five of five, and a retraction of a retraction |
| 10 | the byte that is not a timezone |
| 11 | two catalogues disagree; what the object can settle and what it cannot |
| 12 | 95.8047 %, 75.2786 % and 19.7139 %: three numbers, three denominators |
| 13 | three dates on the disc and none of them a chain of custody |
| 14 | 1,314 hashes against 77 repositories: zero, and a gap that is still a gap |
| 15 | nine candidates: four solved, three unexplained, two the same thing |
| 16 | nine errors in the pre-briefing, six of mine, one owed to the session before |
| 17 | what is still open, and what it would cost |
| 18 | the predictions, scored: inherited 21.0 / 22, open 53.5 / 71 |
1,208 AVI files on a platform where AVI was believed unsupported. They are
genuine Video for Windows 1.0, BI_RLE8, two streams each, and the disc
registers the extension with MCI in its own WIN.INI and tunes the player.
That establishes what somebody shipped, not what the ROM does — and the
two non-standard AVI drivers this disc names, \mci\coreavi\mciavi.cor and
\mci\modavi\mciavi.mod, are not on it. Chapter 08
stops at that edge deliberately.
A retail title imports HC.DLL. §7 of the platform notes says, from
homebrew, that the hand-controller library must be statically imported and that
its presence in an import table detects a title that reads the controller.
RTC.EXE's only import by name, out of 89 relocations, is HC.HCCONTROL.
CONTROL.TAT is 463 bytes plus its program list, on five discs of five.
The previous session in this pipeline published a correction to the platform
notes on a sample of four. It is withdrawn here, and replaced: the notes'
table put two different fields in one column, and once they are separated
every one of the five discs' total lengths falls out of one subtraction.
80.2861 % of the disc is copies of itself. Sixty filmed clips shipped nine to eleven times each, forty drawn boards shipped twice each. The other VIS pressing shipped its interface art three times over — 1.1024 % — and the platform notes called that a design signature. On this disc it is the disc.
The pictures were decoded and looked at before the percentage was written. They are photographs: a person in a park performing sixty actions, and a pair of hands holding a Memorex controller. The prediction said drawn animation, named a mechanism, and was wrong; chapter 12 has the three statistics, the control population, and the loss.
370 Python files inherited, and six touched or written here:
ne.py |
extended — imports by name from the relocation records, RT_STRING decoding, resource dumping, name-string type ids, and a --validate mode with two negative controls |
avirle.py |
new — a BI_RLE8 decoder for Video for Windows 1.0, with --validate, --census, --flatness, single-frame PNG and contact sheets. It descends into LIST rec because a reader that does not sees zero frames in every file |
tiles.py |
new — reassembles 80 × 60 tiles into the picture they were cut from and scores the result against a shuffled control, saying out loud when nothing separates |
inimods.py |
new — every module a Windows .INI names, checked one at a time against a tree, printing a line for present and absent alike |
tzfield.py |
new — twelve candidate readings of ISO 9660's timezone byte, each as a count out of the population, plus the cardinality test that settles it |
tatlen.py |
new — the five-disc CONTROL.TAT length model |
Every derived byte went to _work\, which held 754,254,358 bytes in 3,696
files when it was measured, and was then deleted. Nothing was executed; every format was read
against its public definition, named where used, and validated on a specimen
that had to fail.
- vis-platformnotes-doc — the shared checklist. This disc touches eight of its sections and adds two things it had never seen
- vis-sherlockholmes-doc — the other documented VIS pressing, thirteen days earlier, same
Maketatbuild - vis-gamelist-doc — the index