Optimize TryLibrary.lua

[Validark]

• Use __index metamethod instead of replicating pointers for each table
• Remove unnecessary Data and TargetError tables and instead only use one table and table.remove
• Removed unnecessary tables
• Removed unnecessary unpack calls
• Replaced Argument parameter with self (more OOP style)
• Allow for callable tables to be passed as the parameters Function and Callback
• Reduce unnecessary/duplicate function calls
• Make Retry have uniform logic after resetting attempt state
• Make __tostring metamethod include an id
• Switch ipairs and pairs out for simple for loops (which are much faster)
• Add description
• Add readme link to top of file
• Remove _IsAttempt (which is writable to any table) and instead identifies Attempts with getmetatable(self) == Attempt
• Added pcall Packaging internal helper function which is faster than table.remove and doesn't require the creation of more than a single table

And, it's shorter :D

Note: Didn't change actual functionality

[Validark]

I've integrated this module into RoStrap, which is available here, meaning users can install it directly using the plugin. If F3X requests that it be removed, I will remove it.

[Validark]

Feedback? Suggestions?

[GigsD4X]

This is great! Just got back from finishing a massive project over the past 2 weeks, will review these changes tomorrow 👍

One issue might be that comparing the metatables of potential attempt objects to Attempt could make attempts generated by different instances of the library incompatible. If the goal is to avoid a false positive when _IsAttempt happens to be in a returned, non-attempt table, perhaps we could move _IsAttempt into the metatable itself where it is less likely to conflict?

[Validark]

What scenario are you envisioning where a user generates an Attempt via Tryfrom the library, then calls the Attempt methods on that Attempt from a copy of the library or another version?

This just doesn't make sense to me, I don't see how that could be an issue. If people are using the prescribed method syntax, even if a user is for whatever stupid reason running 5 of the exact same library, there wouldn't be a conflict.

[GigsD4X]

This is possible in cases where an arbitrary system might come prepackaged with the library as a dependency in order to generate and return attempts that are intended to be used by any other arbitrary script (which may use a separate copy of the library to handle that returned attempt)

[Validark]

Again, still impossible if a person uses method syntax

[GigsD4X]

Merged these changes in 736af3be622eade2b0e7034d731f79c19f87fe08 for 2.0.0 🎉