feat: live reloading when running as a module

[VishnuSanal]

Description

This PR fixes #654

how to run: python3 -m app --dev

[vercel]

@VishnuSanal is attempting to deploy a commit to the sparckles Team on Vercel.

A member of the Team first needs to authorize it.

[codspeed-hq]

CodSpeed Performance Report

Merging #858 will not alter performance

_{Comparing VishnuSanal:module-hot-reload (a7272e5) with main (1e33769)}

Summary

✅ 109 untouched benchmarks

[sansyrox]

Hey @VishnuSanal 👋

I am unable to understand the changes. Could you add an example on why we need the new parameters?

And how you'd invoke the changes?

[sansyrox]

Hey @VishnuSanal 👋

I am unable to understand the changes. Could you add an example on why we need the new parameters?

And how you'd invoke the changes?

Hey @VishnuSanal ,
Could you please update the PR description with the above??

[VishnuSanal]

uh, oh. my previous comment never got sent.

Could you add an example on why we need the new parameters?

I have changed the implementation, PTAL. removed the need for new params.

And how you'd invoke the changes?

running as a module usually, like python3 -m app --dev would do.

please CMIIW.

[sansyrox]

@VishnuSanal , I just saw your description. That is not what we want to solve for 😄

We want to allow the dev mode to work like robyn -m src --dev. The robyn_app in description is src here

where src

src/
  - __init__.py
  - __main__.py

The --dev flag should rerender based on app changes.

[VishnuSanal]

thank you for the review, Sanskar! this currently works as expected in my testing, PTAL & LMK what you think. :)

[sansyrox]

Hey @VishnuSanal 👋

Thanks for the PR, I have some comments 😄

[sansyrox]

Hey @VishnuSanal 👋

Thanks for the update. I have some follow up comments

[sansyrox]

Can we not use self.file_path?

Then we don't need to pass in config.

[VishnuSanal]

we also need config.dev. should I pass it as a variable?

edit: we also need running_as_module now.

[VishnuSanal]

current issues:

• doesn't work when using robyn -m src --dev. sys.argv[0] returns the path to the executable. but, from here, it seems like we shouldn't run it like it, but like python3 -m src --dev.
• doesn't work during python3 app.py --dev (since path is not present in unknown_args in arguement_parser)
• doesn't work during robyn app.py --dev

(will edit & update this comment as I fix them...)

[VishnuSanal]

note to self:

the flow of control is like: cli#run => cli#start_dev_server => reloader#setup_reloader => EventHandler#reload => Robyn#start and, we are getting the file object at the last stage. my current implementation starts a new dev server from there.

we need a way to bypass it the first time (inorder to reach robyn#start) and, then kill the previous reloader & start new one from there.

as of now, I am getting an infinite loop! will address. :)

[recurseml]

⚠️ Only 5 files will be analyzed due to processing limits.

[recurseml]

Unsafe access to sys.argv[0] without checking if sys.argv is empty. This can raise an IndexError when Python is run in environments where sys.argv is an empty list (e.g., certain Python interpreters or when running through specific APIs). The code should check sys.argv length before accessing the first element. A safer approach would be: self.file_path = sys.argv[0] if len(sys.argv) > 0 else ''

📚 Relevant Docs

---

React with 👍 to tell me that this comment was useful, or 👎 if not (and I'll stop posting more comments like this in the future)

[recurseml]

The import changes introduce a breaking change in dependency initialization. The old code imported and used setup_reloader directly from reloader.py, but the new dev_server.py only imports the reloader module without importing setup_reloader. This will cause a NameError when the dev server tries to use setup_reloader since it's not properly imported in dev_server.py.

---

React with 👍 to tell me that this comment was useful, or 👎 if not (and I'll stop posting more comments like this in the future)

[recurseml]

Unsafe path resolution: Using resolve() on a potentially non-existent path without error handling can raise FileNotFoundError or other path-related exceptions. This could crash the server when given invalid paths. Should wrap the path resolution in a try-catch block to handle potential path-related exceptions gracefully.

📚 Relevant Docs

---

React with 👍 to tell me that this comment was useful, or 👎 if not (and I'll stop posting more comments like this in the future)

[recurseml]

😱 Found 3 issues. Time to roll up your sleeves! 😱