Hacker News Reader

Comment by jamesgeck0 | Hacker News Reader

Still, Chrome's updates tend to be measured in kilobytes, and the update for Firefox 5 was a few megabytes. There's room to shrink things.

View on HN

This is because Chrome has a little disassembler built into their binary-diff mechanism, so they can get really small diffs (at least on Windows; for Linux they fall back onto bsdiff).

http://dev.chromium.org/developers/design-documents/software... has some of the details. Note the ~90% reduction in size going from bsdiff to this.

I want to use this for my own stuff, because that's cool as hell, but the Courgette source has pretty major dependencies on the rest of Chromium.

Doesn't mean it's necessary. The size of the update matters little to the end user as long as it's small and 5 megs is small for most people.

The update server was delivering the ~8mb patch at about 4kb/s this afternoon. It took me several minutes to update.

Even when update servers aren't dying under heavy load, my internet connection isn't very fast. There are many people who don't have fast connections, both in the US and elsewhere. Programs downloading things in the background render web browsing noticeably slower and make voip unusable. Update size matters.

It matters for some people, yes, like you, but for most, it simply doesn't. I did say most in my original post as well.

Glancing at stats from Point Topic[1] and Internet World Stats[2], it looks like only a quarter of the world's internet connections were broadband in 2009. If one is building software for a global user base, a fast connection is not a given for most people.

1. (pdf) http://broadband.cti.gr/el/download/World%20Broadband%20Stat...

2. http://en.wikipedia.org/wiki/List_of_countries_by_number_of_...

Reply on news.ycombinator.com