Jeff Smith wrote:
> By "valid timestamp", I really meant a valid timestamp of last
> modification.
Last modification of what? The file name? Its location? It's storage
format?
> In many filesystems there certainly is another
> timestamp which does not change when modified, but the modified one
> is invalid if it does not show when last modified. Well, there should
> be no argument there, but it should be obvious why a "mod" timestamp
> is fundamental to knowing when a file has been changed to it can be
> archived (hence the invention of version control). Think about the
> original purpose of the "archive attribute" in DOS. It was meant as a
> fast way to see what files had changed for the purpose of archiving.
> It became useless when not every editor took care to set it when
> modifying the file.
The problem is not likely to be something that makes a modification
without changing the timestamp but rather the opposite: something that
tries to preserve what it thinks is the correct timestamp while moving
the content around - zip, tar, and an assortment of other ways of
copying files will do that. But consider what happens if you zip a copy
of something, change/commit the original, then unzip your old copy back
to undo a mistake. It's now a change as far as svn is concerned but the
file has an old timestamp. Or, you might copy some files out to work on
a different machine where the clock setting is wrong, than copy back
with a method that preserves that wrong timestamp.
--
Les Mikesell
lesmikesell@gmail.com
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@subversion.tigris.org
For additional commands, e-mail: users-help@subversion.tigris.org
Received on Wed Feb 14 19:32:03 2007