Repository navigation
py.RunFile / RunCode doesn't handle a nil module pointer #249
Description
Hey, there. It's really cool to be able to run Python in Go! I'd like to congratulate you on this project, it might be exactly what I've been looking for when it comes to an easily embeddable scripting language.
I just noticed that py.RunFile() and py.RunCode() don't handle a nil py.Module pointer when deciding to create a new module. I feel like that would lessen the friction on using it, since if you run multiple files and want them to be in a single module you currently have to call two different RunFile() / RunCode() calls that are essentially the same (just with nil as the inModule argument the first time).
Basically, currently I'm doing this:
var mod *py.Module
...
// In a for loop for mutliple .py files
if mod == nil {
mod, err = py.RunFile(context, fpath, py.CompileOpts{UseSysPaths: true}, nil) // Returns a new module
} else {
mod, err = py.RunFile(context, fpath, py.CompileOpts{UseSysPaths: true}, mod) // Adds onto the existing one
}and it would be easier to just do this:
var mod *py.Module
...
mod, err = py.RunFile(context, fpath, py.CompileOpts{UseSysPaths: true}, mod) // Returns a new module on first run because 'mod' is nil, adds onto the existing module afterwardsI believe to implement this I would just need to modify RunCode() to check for nil in the case *Module part of the inModule type check. Does this sound OK?
Activity
ncw commentedon Oct 23, 2025
Your before and after seem to be the same so I'm not sure I understand!
Is exactly the same as this isn't it as
modwill benilon first run.SolarLune commentedon Oct 23, 2025
Hello - the difference is that I'd like to not call
py.RunFiletwice, just once. Unless I'm misunderstanding, I have to call it twice currently because passingnilin as an argument to create a new module isn't the same as passing a nil pointer to a *py.Module, as this part ofRunCode()switches off of the argument type explicitly:gpython/py/run.go
Line 154 in e20a7a4
So running it with
nilforinModulecreates a new module, and running it with a pointer to a module forpy.Modulewill update the module, but passing a nilpy.Modulepointer will cause a crash.drew-512 commentedon Nov 10, 2026
Confirmed on main (444ae5e): a typed nil
*py.Modulepassed toRunSrc,RunFileorRunCodepanics with a nil pointer dereference onmod.Globals, because a nil*Moduleinside aninterface{}matchescase *Module, notcase nil.#260 fixes it the way suggested here: a nil
*Modulecreates a new__main__module. Its test panics on main and passes on its branch, and the loop from this issue (mod, err = py.RunFile(..., mod)) works with it. It also merges cleanly alongside #258, which touches the same file. #260 shows no CI run yet; GitHub holds workflow runs on a first-time contributor's PR until a maintainer approves them.