[darcs-users] Re: peer to peer darcs

[Eric S. Johansson]

David Roundy wrote:

> It all depends on the size of your repository (and of course, how many
> patches are being left out of your "minimal" set.  On my computer, a darcs
> get of the linux kernel repository (20,000 changes) takes about a day, just
> to apply the patches.  I'd (very roughly) expect the same time for creating
> a minimal patch bundle when there are somewhere around 100 patches involved
> (assuming they are the size of linux kernel patches).
> 
> Another *major* issue is memory use.  On any reasonably large repository
> you can't hold a reasonable fraction of the patches in memory.  In the
> existing push and pull code, you already have to hold in memory all the
> patches being either pushed or pulled, plus the patches that are being
> "skipped".
...

[[topic drift]]

now I understanding more clearly why some people have concerns that the 
number of patches will increase to the point where you can not 
reasonably manage them.

If catching up with patches takes more than 15 minutes, I would probably 
be terribly concerned that something was going wrong.  Is there some 
sort of mechanism to collapse the history into new reference point from 
which one could start applying a subset of the patches.  This is not to 
say that one should delete the patches previous to the reference point 
because that used to your history but only it's the reference mark from 
which you construct the rest of the history.  One potentially convenient 
reference point creation time would be the tagging process.

make sense or am I way off base with my understanding?

---eric