|  | Commit message (Collapse) | Author | Age | 
|---|
| ... |  | 
| | |  | 
| | 
| 
| 
| | Found via behaviour of pushnumint in lmathlib | 
| | |  | 
| | |  | 
| | |  | 
| | |  | 
| | |  | 
| | |  | 
| | |  | 
| | |  | 
| | |  | 
| | |  | 
| | 
| 
| 
| | Fixes `0.983 == 0` == true | 
| | |  | 
| |\ |  | 
| | | |  | 
| | | |  | 
| | | |  | 
| | | |  | 
| | | |  | 
| |/  
|   
|   
|   
| | If there's a table in the concat sequence, we ended up with a toconcat
array with undefined slots. | 
| | |  | 
| | 
| 
| 
| | Closes #37 | 
| | |  | 
| | |  | 
| | |  | 
| | |  | 
| | |  | 
| | |  | 
| | |  | 
| | |  | 
| | 
| 
| 
| | Do the TString creation in one big Array.prototype.concat | 
| | |  | 
| | |  | 
| | |  | 
| | |  | 
| |\ |  | 
| | | 
| | 
| | 
| | 
| | | Oddly 'const' on OCi helps, but on anything else dramatically increases runtime
Otherwise the reduced scope for other variables is a tiny bit faster | 
| | | 
| | 
| | 
| | | This makes it faster on my computer | 
| |/ |  | 
| | |  | 
| | 
| 
| 
| 
| 
| 
| | No longer follows C api as closely (it is the way it is because C has unions)
Instead, have a flat object.
Also remove the 'setval' method, as it wasn't used from all call-sites and couldn't be extended to them (setval's body only worked with source TValue on a stack) | 
| | |  | 
| | |  | 
| | |  | 
| | |  | 
| | |  | 
| | |  | 
| | |  | 
| | |  |