Comment author: @rversteegen

(Noooo!!! I was scared this would happen somehow)

"This means that loading any version 5 RPG in the serendipity version of game.exe
will result in broken shops."

--------
DIM b(curbinsize(1) * 25)
recordsize = getbinsize(1) / 2
...
setpicstuf b(), recordsize * 100, -1
loadset game$ + ".stf" + CHR$(0), id, 0
--------
where b() is called like "b(o * recordsize + i)"

This code looked right at the time, but now that I look at it, read data with an
i larger than recordsize would read into the next shop item's data, and all the
real uninitialised data is at the end of b(). I suppose the shop stuff then
needs to be spread out over b() and recordsize set to curbinsize(1)/2

What I don't understand is why I didn't run into this when testing unupgraded
copies of TEST and WANDER. And also, why it took over a week for anybody to
notice this when it should break every shop in every game not released after S.
or edited by the person playing them.

EOTE2 is such an old game, and probably hasn't been updated since the author
disappeared many years ago, so it's very likely version 4 or less.

Comment author: @bob-the-hamster

Yay! I solved this one. Because each .STF record contains multiple sub-records
for each item in the shop, resizing this lump requires some extra code. The meat
of the fix is wrapped in this new loadshopstuf SUB

SUB loadshopstuf (array(), id)

ol = getbinsize(1) / 2 'old size on disk

nw = curbinsize(1) / 2 'new size in memory

flusharray array(), nw, 0

'load shop data from STF lump

setpicstuf buffer(), ol * 2 * 50, -1

loadset game$ + ".stf" + CHR$(0), id, 0

'in case shop data has been resized, scale records to new size

FOR i = 0 TO ol - 1

FOR o = 0 to 49

array(o * nw + i) = buffer(o * ol + i)

NEXT o

NEXT i

END SUB

check subversion for the full fix.