Tzvetan Mikov
Tzvetan Mikov
To be clear, I am not insisting that this be included, but I feel that it is a more common use case than is commonly assumed, so I am just...
@michaelficarra for the record, we do care about conformance a lot. That is precisely why I am filing this question and why we are attending TC39 meetings. My point is...
Wow, I don't think we have ever received such thoughtful and high quality compiler PR. This will be a meaningful improvement to Hermes, plus it starts the work on normalization,...
> For introducing the NormalizeInst pass, my only question is at what point in the optimization pipeline should it run? I agree that it makes sense to start by inserting...
FYI, we just enabled the "Discussions" tab, for more informal questions and ideas.
Hi, sorry we haven't commented here in a while,I haven't forgotten about these PRs, though admittedly https://github.com/facebook/hermes/pull/1175 made me feel more guilty. Since we have now publicly announced Static Hermes,...
Thank you for this bug report. It is an interesting case. In the first iterations of the loop, the situation is like this: you have an element at index 1,666,535...
To further clarify the current Hermes behavior. It allows us to support cases like this, which would normally run out of memory trying to allocate many gigabytes of data: ```...
FWIW, we have a plan for fixing this. But everyone, please keep in mind that creating and using sparse arrays is almost never a good idea
`(new Array(6666145 >> 2)).fill(0)` creates a dense array. It is pretty efficient.