Re: svn commit: r1102690 - in /subversion/trunk/subversion: libsvn_ra_serf/replay.c tests/cmdline/svnsync_tests.py

On 05/13/2011 03:05 PM, Kamesh Jayachandran wrote:
>
>>Yeah, Kamesh is just incorrect on this. The editor interface has always
>>allowed multiple simultaneous open files to support what it calls "postfix
>>text-deltas".
>
> I could not understand that. I will try to understand that and fix my code.

[For the record, I'm not saying your code is wrong -- I didn't even look at
it. I'm only correcting your statement about simultaneous open files.]

Because transmitting file contents is generally considered the most costly
part of describing a tree delta, the editor interface allows you to describe
all the "tree changes" and property changes before transmitting any file
content changes. (This allows editor implementations to detect
out-of-dateness and conflicts before the file content changes are
transmitted). So, for example, the following is valid:

   # transmit tree structure changes
   d1 = open_root()
     d2 = open_dir('A')
       f1 = open_file('A/mu')
       d3 = open_dir('A/B')
         f2 = open_file('A/B/alpha')
         f3 = open_file('A/B/beta')
       close_dir(d3)
     close_dir(d2)
   close_dir(d1)

   # transmit post-fix text deltas
   apply_textdeltas(f1)
   close_file(f1)
   apply_textdeltas(f2)
   close_file(f2)
   apply_textdeltas(f3)
   close_file(f3)

   # complete the editor drive
   close_edit()

Naturally, in such situations, you don't want to open your file batons in
pools that will be destroyed when their parent directory baton's pool is
also destroyed. You need instead for file baton's to have lifetimes that's
about as long as the whole editor drive.

-- 
C. Michael Pilato <cmpilato_at_collab.net>
CollabNet   <>   www.collab.net   <>   Distributed Development On Demand