2010/6/29 Andy Levy <andy.levy_at_gmail.com>:
> On Tue, Jun 29, 2010 at 09:37, Ralf Onat <kongomongo_at_gmx.net> wrote:
>>> >> Subversion doesn't store the size of each file as metadata, so to do
>>> >> this one would have to run svn cat against each file's URL for each
>>> >> revision & count the bytes. To do so would be extremely expensive.
>>> > I guess it's unlikely that this change can be pushed into subversion then? What would be your guess?
>>> I would guess that it'd be a very hard argument to win. Such a feature
>>> would hammer disk & network (since you can't stream a diff to the
>>> client, each operation would require the full file) I/O & have the
>>> potential to be very easily abused.
>> Err. Sorry I was referring to *SUBVERSION* storing the file size within the metadata! So TSVN can just display the size without any additional overhead.
> Similar answer. When Subversion stores each revision in the
> repository, it often will only store the diff. So to record this, it
> would have to internally "svn cat" the whole file to get a file size.
> On a large commit, this could be very expensive. I'm not sure that the
> commit process could handle calculating it on the client end &
> transmitting it as part of the commit; there may be some huge holes in
> that idea which I can't think of offhand.
Because of svn:eol-style handling the file size will be different on
different platforms. There is also svn:keywords.
And I do not think that the feature asked by the OP will be useful.
To unsubscribe from this discussion, e-mail: [users-unsubscribe_at_tortoisesvn.tigris.org].
Received on 2010-06-29 16:02:59 CEST