Optimize session deletion, recreation, and persistence

[callmetwan]

This work optimizes control flow by only rotating the session key or persisting data if the session is not deleted. The flow before this work would recreate the session key/persist session data every time, even if it would end up being deleted. This is important if storing the session in a database, as that means there is at least one additional round trip. For larger systems this could mean cutting the number of database interactions in half.

Tangential, the bun dev script has been split into two to reflect the two types of servers you can run.

Before merging I would love it if someone else could test this as well. Not being overly familiar with the library and there being no unit tests makes me nervous I missed something 😅

[callmetwan]

The original conditionals were logical but a bit challenging for me to grok. I'm a big fan of abstracting conditionals into well named variables. I personally lean towards verbosity over succinctness to make the intention really clear.

That being said, happy to make changes if there is different stylistic preference on how to write these conditionals.

[callmetwan]

A few other notes:

• There wasn't a style guide. I used prettier to format just my changes. Happy to turn use another formatting tool, I just didn't see anything required/referenced in the repo
• I lean towards verbosity over succinctness in variable naming to very clearly illustrate intent.
• I left larger docblock style comments to also illustrate intent. I specifically leave out variable name references in comments so the comment doesn't atrophy/make the code reading experience confusing down the line.

[callmetwan]

None of these changes are intended to alter the functionality, just re-organize the code so we aren't unnecessarily updating the session store if we don't need to.

[jcs224]

Looks like something broke the cookie driver.

[callmetwan]

I'll take a look later this week

[callmetwan]

Hmmm, running it locally all the tests pass. Here is the node cookie test for example

It seems to be failing on the incorrect password step but I verified in Chrome and FF (on a Mac) that they work as expected. Is there a way to re-run the tests in GH actions? I looked but didn't see any options.

[jcs224]

Weird. Did you change the STORE environment variable to cookie before running the test?

I'll run the action again.

[jcs224]

Still failing, I'll check this out in the next couple days.

[callmetwan]

Ah, I didn’t realize I needed to pass an encryption key. I’ll remind myself to add this to the testing section of the README

[callmetwan]

Okay, think I fixed it! Had a ! in the wrong place

[jcs224]

Nice! I might play with it a little before I merge it in, but this looks good, thanks!

[callmetwan]

Please do, I want to be 100% sure I didn't mess anything up unintentionally! And glad to do it :)