>>78122536>>78122645Seems bison 1.29 did the trick, there's even a win32 build of it available.
Had to modify the output file though, it looks like MS themselves customised bison's skeleton file, which I picked up on from
>Tail of namespace wrap is located in bison.skl to allow for wrapping>// auto-generated tables>//} // namespace DirectUIlines in parse.y, and then an error in the build about namespace not being closed, just added an extra } somewhere and it seemed to work.
The other error it gave is kinda weird, it's because of
>int yylex(BOOL* pfRes);in the parse.y, so not bison's fault really.
yylex gets defined in scan.cpp (which flex parses from scan.l, luckily we have prebuilt scan.cpp though), and there's a comment:
>// DirectUI: yylex (YY_DECL), out param used as success code (yyterminate() called on failure)(this comment isn't in the scan.l though, so it must have been part of flex's skeleton file)
Removing the BOOL* pfRes from parse.y/cpp seems to let everything build, but I'm not sure how well it'll run like that.
Pretty sure using wrong params when calling a function is a bad thing, and probably should handle that pfRes result too, if it was important enough for them to change shit for.