Initialize bytefmt package #1
Conversation
| } | ||
| } | ||
|
|
||
| func assertNoErr(t *testing.T, err error, message string, args ...interface{}) bool { |
There was a problem hiding this comment.
These are here because this library is intended to be a common dependency. I don't want to pull in testify for such a small test suite.
There was a problem hiding this comment.
Nice interface. I like it.
Still not sure which of the tests would fail in the other libs that do similar things, so a comment there would be helpful to motivate this to someone who isn't clued in to the world of byte-formatting libs. Not as a jab to those libs, but just to be transparent about it.
This should be open-source, there's nothing proprietary or secret here.
| } | ||
|
|
||
| // To avoid precision loss for large numbers, calculate size in big decimal. | ||
| // value = (whole * 10**len(frac) + frac) * scale / 10**len(frac) |
There was a problem hiding this comment.
It's a summarized version of the arbitrary-precision math we're doing below. I'll throw in an example, something like the following:
value = (whole * 10**len(frac) + frac) * scale / 10**len(frac)
123.456 GiB = (123 * 1000 + 456 ) * (1024 * 1024 * 1024) / 1000
= 132559870623
| func (s *Size) Add(y Size) { s.bytes += y.bytes } | ||
|
|
||
| // Sub subtracts size y from the current value. | ||
| func (s *Size) Sub(y Size) { s.bytes -= y.bytes } |
| // -1 if s < y | ||
| // 0 if s == y | ||
| // +1 if s > y | ||
| func (s *Size) Cmp(y Size) int { |
There was a problem hiding this comment.
Is Cmp used anywhere in the standard library? This looks very similar to strings.Compare and bytes.Compare. I think it would be clearer to spell it out unless this is commonly used.
There was a problem hiding this comment.
These three-letter abbreviations look to be idiomatic in math packages (big.Int, Shopspring's decimal.Decimal and Kubernetes' resource.Quantity). My preference would be to err toward consistency with those packages since this lies within the same domain, but I'm open to a counter-argument.
This package solves a number of issues we've encountered with byte quantity parsing and formatting. One or more of the following problems appear in every existing library I could find.
This package is inspired by Kubernetes' Quantity value. In most cases it's not appropriate to use that package because it requires a dependency on Kubernetes. It's also much more awkward to use due to being a generalized solution that does not presume byte units.
Remaining work: