WIP: Experimental support for integers as JS bigints #404
Conversation
There was a problem hiding this comment.
We will try.
How deep is the rabbit hole? :) This can be told by the developers of sbcl, who have 70% code of the NUMBERS this is the difference between integer and bigint.
There are algorithmic features and differences for integer and bigint as well as when working with bits ops.
Adding type checking to a very usable predicate integerp will degrade performance highly likely - It will be necessary to watch.
But probably this should be done in a separate release or use conditional compilation.
bigint implementation in JS is very stranged. :)
And not the last question - how many users this features is expected to have? This is the first question on the topic in 4 years :))))
We will try :)
Agreed! I added a I don't know yet how far is the rabbit hole :-) still fixing errors. I think I'll try to get it to boot a bit more as an experiment, trying to make it conditional, and re-evaluate the strategy later. |
|
The question is - will we enter the data type "0n123" for reading by the reader ? |
No. My plan was to read all integers as bigint (if no decimals point is present) and float otherwise. |
|
When do we expect a release? |
Do you think it would be valuable? I can definitely spend a bit more time on it. |
|
I don't see how to use it. There are not many people who are really involved in this area. And they have other tools. |
|
I'm considering issue #420. |
Just an experiment.
This is work-in-progress and the PR does not work yet.