On 26 February 2012 16:05, Stefan Sperling <s...@elego.de> wrote: > On Sun, Feb 26, 2012 at 03:51:44PM +0000, sebb wrote: >> Now release archives can be quite large, so one wants to minimise >> network transfers, and not have to checkout large workspaces either. > > Why would anyone need to get a working copy of anything? > All moves/copies can be performed server-side. > The RM can import the file into a directory, and others can download it. > > Because separate revisions (RCs) of the same release usually use unique > filenames, there is currently no benefit in updating an existing working > copy instead of downloading the release artefact. > >> In this case, there is no parent directory which can easily be >> renamed, so what are the options for minising network usage? > > What network usage are you trying to minimise? > RM to SVN server?
Yes. > Other developers to/from SVN server? No, they would have to download just once from the dev area. > SVN server to release mirror servers? No. >> One could rename each file individually on the server, but that would >> result in lots of separate commits, and would be tedious and error >> prone. > > In this scenario, I have no better idea than renaming files individually. > >> Is there a way to rename multiple files as part of a single commit? > > Yes, there is. See svnmucc: > https://svn.apache.org/repos/asf/subversion/trunk/tools/client-side/svnmucc Is this available for all OSes? Is it documented anywhere? Whatever process is chosen needs to be easy to use for multiple RMs.
