KIO and KFM and copy/move/delete operations

[Rajko Albrecht]

On Tuesday 18 October 2005 01:47, Thiago Macieira wrote:
> A simple operation is on one file only, or remote-to-remote. In that case,
> I see no difference: the ioslave queues up the requested URLs and then
> sends one command when the commit() function is called. Just as if you
> had a checkout.
Once again: you have NO commit on remote. This is just like working on a 
special filesystem. "Commit" is just for working from a (special managed) 
local working copy and means nothing more than "sync my local filesystem to 
remote filesystem". 

Branches and Tags you'll make on repository 'cause it is much faster than 
creating them on local working copy and commiting them. That means a copy 
operation on a whole directory. eg 

"svn copy <svn-prot>://<host>/<path-to-repo>/trunk 
<svn-prot>://<host>/<path-to-repo>/branches/my-cool-branch"

With kdesvn and svn it is a short-time operation, cost nothing on serverside, 
no space (just a few bytes) etc. With KIO due that recursive mkdir/get/copy 
of each single item it cost a lot more depending of the tree. And exactly for 
such stuff users want KIO so them can make such simple operations from within 
there filemanager and not needing a special client. And exactly that is not 
useable at moment. Creating them local means a that you have to checkout the 
WHOLE repository, eg, your main and branches and/or tag path at once and copy 
then. That means - you have a complete copy of repository. Take a look into 
kde-subversion server and just think about getting all 
within /branches, /tags and /trunk at once. Exactly thats the reason for the 
idea making branches/tags as a simple server operation.

On other hand it makes real no sense (of course just IMHO) that kfm will not 
use capabilities of a specific protocol which will let operations much more 
faster. Of course it should make all that checks on "simple" protcols like 
smb or ftp. But why not use when a transaction based server like subversion 
may do that check by them self? Why not just asking the protocol "Do you want 
handle recursive copies and moves yourself?" I think this is a decision the 
specific KIO and/or user should make not KFM. And sorry: checking for such a 
boolean costs definite less time and resources than copy each single file 
alone within svn.

cu
Rajko
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: not available
URL: <https://mail.kde.org/mailman/private/kfm-devel/attachments/20051018/3728a9b4/attachment.sig>