Second update clones entire firmware history

[fastcat]

The first run of rpi-update does a git shallow clone, which is nice & appropriate.

However, the second run does a "normal" fetch, which pulls in the entire history, which is hundreds of MB and takes quite a long time, both for github to prepare and to transmit across the network.

Using git fetch --depth=1 pulls in more than would be downloaded if a fresh shallow clone was done, but vastly less than a full fetch does.

Given that even a shallow fetch downloads about double what a fresh shallow clone does (for only a handful or commits needing to be fetched), it seems like always doing a fresh shallow clone would be more efficient.

[popcornmix]

Interesting.

[Hexxeh]

This would explain why things have become painfully slow!

[popcornmix]

I've tried the git fetch --depth=1 and it works for me, so I've pushed it.
@Churten thanks for bug report.