[myself248]

What gets me is that ZModem-Resume had effective continuation of failed transfers _decades_ before HTTP got it right. The early days of the web drove me bonkers, having to start over with a long file that failed in the middle.

[Sohcahtoa82]

I remember using a program...I forget what it was called, but it was popular in the 56K days for resuming HTTP downloads. It would even allow you to provide mirror URLs so you could download from multiple sources at once.

I want to say it was called WinGet, but now there's something else called that, so I can't confirm it.

EDIT: It was called GetRight.

[blueflow]

> before HTTP got it right

In the spec, yes. But since servers use Content-Encoding instead of Transfer-Encoding for on-the-fly compression, its unspecified now whether the byte offsets in the resumption refer to the compressed or uncompressed file. Result: Download resumption did not work reliably and got removed from browsers.