add MustFromBig for nicer struct initialization

[karalabe]

This PR allows the second form (last two lines) along side the current form (first two lines) of initializing new ints from bigs.

var (
	costCap, _   = uint256.FromBig(item.Tx.Cost())
	gasTipCap, _ = uint256.FromBig(item.Tx.GasTipCap())
)
meta := &blobTxMeta{
	id:            id,
	size:          size,
	nonce:         item.Tx.Nonce(),
	costCap:       costCap,
	gasTipCap:     gasTipCap,
	gasFeeCap:     uint256.MustFromBig(item.Tx.GasFeeCap()),
	blobGasFeeCap: uint256.MustFromBig(item.Tx.BlobGasFeeCap()),
}

Panics can only be captured when a goroutine is unwinding, hence why the messy construct in the tests. Also, it would have been a tad cleaner to wait on the panic with sync.WaitGroup vs a channel, but the channel approach introducing a dependency on sync.

[karalabe]

Actually, might have been nicer to just call FromBig and handle the return, more future proof.

[holiman]

Actually, might have been nicer to just call FromBig and handle the return, more future proof.

You mean in this PR or from your side?

[holiman]

Actually, I'm thinking that if you're in this case:

gasFeeCap:     uint256.MustFromBig(item.Tx.GasFeeCap()),

And let's say we know nothing about item.Tx at this point. Then it would be better to use a method which silently just overflows, not overflows+panic. Because if this happens early in the stack before the tx is properly validated, it is better to obtain a "wrong sender" than crashing the app.

So wouldn't you rather have a uint256.TryFromBig or something like that?

[codecov]

Codecov Report

Merging #128 (416c76a) into master (c385c61) will not change coverage.
The diff coverage is 100.00%.

@@            Coverage Diff            @@
##            master      #128   +/-   ##
=========================================
  Coverage   100.00%   100.00%           
=========================================
  Files            5         5           
  Lines         1538      1543    +5     
=========================================
+ Hits          1538      1543    +5     

[karalabe]

So wouldn't you rather have a uint256.TryFromBig or something like that?

Not really, all these data are already parsed from disk or network into uint256, only the tx interfaces are still using big.Int and I don't want to overhaul all the interfaces for the blob pool.

[holiman]

LGTM