At 10/5/2005 10:42 AM, Paul Koning wrote:
>You give me too much credit. It only means that I didn't perform the
>experiment that others did.
>However, my comment still applies in this sense: if you're really
>seriously concerned about data integrity, you need to look at every
>part of the chain that handles the data. It may well be that the
>repository has MD5 integrity. But what about the data once it leaves
>there? What data integrity does it have once it lands in your working
>directory?
This (integrity of working directories, particularly on remote client
machines) is a fair a question, yes; and I suspect because of the diversity
of client systems (virtually unlimited), this is difficult to control, to
say the least. One could have the client MD5 files at submission
time...but that won't make any different if the file is already corrupted
before then.
However, for me, this issue is outside the scope of what I'm trying to
address for this line of questioning, and that's the question of wether or
not Subversion 1.2.x and above has it's own "checksum integrity" for *all*
of the content it controls, including all of the file content (be it full
or incremental).
-Matt
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@subversion.tigris.org
For additional commands, e-mail: users-help@subversion.tigris.org
Received on Wed Oct 5 18:14:58 2005