On Sep 16, 2015, at 12:13 PM, Scott Robison <sc...@casaderobison.com> wrote: > > https://www.youtube.com/watch?v=W71BTkUbdqE
The point I took from that that seems most applicable to Fossil is the idea that “workspaces” (i.e. local clones) are not complete copies of the entire repo’s history. In Google’s case, it’s because it would simply be unfeasible, at 86 TB, yet would include millions of old files that don’t even exist at the tip of the trunk any more. While there probably isn’t anyone on this list with a Fossil repo anywhere near this size, the lesson remains: you probably don’t actually need the entire repo history in your local clone. Some months ago, the thread “Fossil 2.1: Scaling” thread touched on this: http://lists.fossil-scm.org:8080/pipermail/fossil-users/2015-March/019952.html In that thread, I proposed that the default mode of “fossil clone” be to include only the last N weeks of work on each open branch. The idea being, you rarely need more than that except for archaeology work, at which point Fossil could go back to the repo it cloned from and import more of the history. The amount of history per branch should be configurable, and of course it should be possible to disable the limit, so that you get a complete clone, which is nice for distributed backups. (I’m reminded of a Linus Torvalds quote, where he said he never did backups, because his code is already distributed all over the world.) Has there been any movement on that front? _______________________________________________ fossil-users mailing list fossil-users@lists.fossil-scm.org http://lists.fossil-scm.org:8080/cgi-bin/mailman/listinfo/fossil-users
