corrected file size to proper use of B vs b and M/G vs m/g

[mhogeweg]

This pull request has been automatically generated by Prose.io.

[benbalter]

To summarize, it looks like the change here is using capital MB and GB to describe the size of datasets/distributions, correct? Could you elaborate on the reasoning behind that a bit, which may help to better inform the decision?

[mhogeweg]

sure: M is for mega (not m), G is for Giga (not g). B is for byte, b is for bit.

[mhogeweg]

check out: http://en.wikipedia.org/wiki/Gigabyte

[konklone]

Well! If we're going to be precise, I must insist we use proper power-of-2 units to represent size. GB is only acceptable to represent literal billions of bytes; in this context, and indeed most contexts when people refer to storage size whether they know it or not, they mean GiB, or Gibibytes. Please refer to Wikipedia's extensive style manual on byte quantity usage for further detail.

👰‍♀️

[waldoj]

FWIW, The Chicago Manual of Style prescribes "MB" and "GB" for the reason Marten cites. (Though it is understandably silent on 1,024 vs. 1,000. :)

[jpmckinney]

Please just state the filesize in bytes. The tiny difference in the JSON file's size between having values like "1000000000" versus "1GB" isn't worth the extremely high likelihood that people will get confused between GB and GiB, etc. Also, it's much easier for an implementer to get a filesize in bytes than it is to transform that filesize to the appropriate abbreviated form.

[mhogeweg]

sure enough. unambiguous units are preferred.

[haleyvandyck]

thanks @mhogeweg for this as well! 🇺🇸