performance analysis: push/pull

Robert Collins robertc at
Thu May 10 08:23:23 BST 2007

On Wed, 2007-05-09 at 15:50 -0500, John Arbash Meinel wrote:
> Hash: SHA1
> Robert Collins wrote:
> > Well, it hurt my head. I think this is reasonable at this point, but I'd
> > love input from $people.
> > 
> > -Rob
> > 
> In my single pass read through, I was able to understand everything. I
> haven't sat down for a full analysis to make sure you covered all edge
> cases.
> Also, I don't think this is all correct reST syntax (:SHA1:, for
> example). reST doesn't care, but it also doesn't do anything with it.

I wasn't expecting anything  :).

> ...
> I'm not sure about having a transmission level hash check (:DELTASHA1:)
> versus having full-text checks. Transmission level checks only let you
> know that your TCP connection works (which has its own CRC +
> retransmission checking), so it doesn't seem to gain much, versus
> knowing that after applying all patches, your text is correct.

I'm not sure about it either. The idea of the transmission level check
was to avoid having to apply patches to know the text is consistent. If
we dont trust the sender then thats clearly not sufficient. I think
Martin nailed the document approach in his commit paper, I'm going to do
another pass over push/pull today.

> Overall, I think it is well written, and certainly worthy of becoming a
> core developer doc.


GPG key available at: <>.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: This is a digitally signed message part
Url : 

More information about the bazaar mailing list