DOpus (80/222)

From:Lee Bosch
Date:11 Mar 2001 at 17:15:50
Subject:[D5] Re: How to copy with clone

Hello Tim

On 11-Mar-01, you wrote:

> _Replying to a message_
>
> By: Lee Bosch <harshness@home.com>
> To: dopus5@lss.com.au <dopus5@lss.com.au>
> On: Sunday, March 11, 2001, 4:15:38 AM
> Re: [D5] Re: How to copy with clone
>
>
> Hi Lee,
>
>>>> A copy should be a copy and not look like a new file. Nyahh!!!
>
> ;-)
>
>> To my thinking, the time stamp is an attribute of the file, not the
>> directory in which it resides. The FileInfoBlock certainly suggests
>> this relationship.
>
> The file date stamps really ought to be relevant to the actual properties
> of the files themselves.
>
> File dates should represent creation or modification dates (which ever's
> appropriate), or both (if supported by the OS). Windows also records the
> date the file was last accessed (which is probably how it manages to
> destroy files that it was only supposed to be "reading").

Window's FAT filesystem doesn't actually support last accessed date; it
only supports last changed date. What individual Windows applications do
instead is bury the creation date in the file and then to resave the file
every time you do something "significant". In many cases, printing a
document is considered a significant change.

> Likewise, directory dates are supposed to represent the last time the
> directory was modified (that's how the Amiga DOS works). If you copy a
> swag of directories, the newly created directories (*), ought to have the
> current date.

This is a sticky one and has been debated much over the years. The fact
that the ReDate program was created was to answer a group that doesn't
share your view. For me, I think it depends on your situation. In the
case of Jack York's question, I think he should be able to preserve the
directory dates. Most of the time, I like using the DOpus date sort to
help me find where the most recent activity has been. It is also notable
that FAT and most Unix filesystems don't update the datestamp on the parent
directory. In Unix, this would be a real problem as you would have to
update all directory stamps up to the root level as each "directory" is
actually a file containing pointers to the files that make up the
directory. "Object Oriented" is *NOT* a new concept.

> (* you are /making/ new directories, regardless of the fact that you're
> "copying" files.)

I believe that the Copy CLONE switch definition is probably the most
"correct" application: Use the old stamp on new directories and the
current date on existing directories.

Lee Bosch



Email majordomo@lss.com.au with 'help' in the body for help.
Members posting binaries to the mailing list will be removed!