A Windows skiing game returned after its creator found the source code he thought was lost
Chris Pirih’s own download page also preserves the unfinished sequel, the old graphics bottleneck, and the limits of what his compatibility notes establish.
SkiFree’s sequel had multiplayer, robot opponents and sound in progress. Then its creator made the skiing stop being fun. That’s a fairly serious problem for a skiing game, even one with other ways to ruin your afternoon.
Chris Pirih’s account on his own site says he began version 2 in 1993 and finished some planned features, including split-keyboard multiplayer, crude robots and sound. The physics became unplayable; losing the original source and turning to other projects then left the released game in what he calls “permanent stasis,” with those sequel features unable to rescue the skiing.
The way out arrived in April 2005, when he found the source for version 1.03 and compiled it into a 32-bit Windows program. He followed with version 1.04 that October, fixing bitmap colors and restoring the slope’s wraparound behavior. The recovery came years ago; the useful discovery now is that the creator’s own page still holds the download and explanation.
Pirih originally wrote the game at home to learn Windows programming. A Windows Entertainment Pack program manager saw it at work and wanted it included. Microsoft licensed the game from him; the skiing experiment became something other people encountered as part of their Windows life.
The game also has an earlier life in text. Pirih describes a Fortran skiing game for VAX/VMS using VT100 escape sequences and scrolling, with trees represented by caret characters. Converting that into a graphical Windows game involved a distinctly less picturesque obstacle: moving pixels over the ISA bus.
He says it was worth substantial preprocessing to reduce video I/O. Combining overlapping objects and erasures allowed smaller screen regions to be copied each frame. The skier got to move smoothly because the computer was being choosy about what it bothered to redraw.
The 2005 rebuild has another practical change. Pirih reports CPU use dropping from 100 percent to one percent, making it friendlier to a notebook battery. That is his historical figure for his rebuild, rather than a fresh benchmark on contemporary hardware. A tiny game needn’t consume the entire processor just because it remembers when processors were tiny.
The page’s malware-warning section deserves attention before clicking past it. Pirih describes scanner alerts, his belief that they are false positives, and additional recompilations from source. His confidence isn’t an independent security finding, and an article cannot turn it into one.
Keeping that context beside the executable makes the page more useful than a mystery download mirror. You get to see what was changed, what was abandoned, and what the author still cannot independently prove about the warnings. Those unresolved alerts remain part of the download decision.
The sequel stayed off the slope.