Re: Issue 730 related question about svn_ra_reporter_t interface

On Mon, 2003-01-27 at 12:30, Karl Fogel wrote:
 Luke Blanshard luke@blanshard.us writes:
  Please forgive me if this is an idiotic question, but can someone
  explain why checkout does anything different from update? Shouldn't
  checkout just create the top .svn directory and then invoke update?

 In essence, what you're proposing is similar to the solutions being
 considered (Ben is detailing them in the issue right now), but not
 exactly the same, because we want to avoid transmitting the entire
 client-side entries list to the server except when absolutely
 necessary.

This answer doesn't seem very satisfying to me. Big updates should also
be interruptable, and once that is taken care of, it should be easy to
reframe checkouts as updates starting from nothing.

I'll put this another way. Consider the following sequence:

  * I issue a checkout for a very large directory.
  * I interrupt it just after it creates the top-level directory.
  * I cd inside the directory and update.
  * I interrupt the update 30% of the way through.
  * I update again.

Does anything inefficient happen? If so, we have a problem with
updates. If not, then why didn't we take Luke's suggestion?

I'm afraid that we may be gearing up to write a bunch of code to the
degenerate case efficiently, when we should be writing a bunch of code
to handle the general case efficiently.

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Sat Oct 14 02:21:31 2006