We need a tool that runs on wikimedia servers, not toolserv. I want to mass move an entire category (pd-old for instance). Toolserv move method creates a duplicate of the image, uses too much bandwidth and hence is not efficient. Images are already on wikipedia servers, no need to involve toolserv.
The assumption is of course that images are not copyvios. I think we should worry less about "what ifs". Eventually all images will have to be reviewed, I prefer that review being done on commons by the 'real' experts.
Of course there is the third question of potential abuse... I think such a tool should be only used by say stewards whom are trusted everywhere. We would make bulk moves from time to time only anyways.
Oh and before anyone asks, if a moved image is not free enough for commons it can always be moved back to the local project - Cool Cat
On 12/14/06, Magnus Manske magnusmanske@googlemail.com wrote:
On 12/14/06, Brianna Laugher brianna.laugher@gmail.com wrote:
On 14/12/06, Magnus Manske magnusmanske@googlemail.com wrote:
On 12/13/06, Bryan Tong Minh bryan.tongminh@gmail.com wrote:
Is there no way you can upload using your own commons login name?
I already thought about extending commonshelper so it accepts commons user name and password. The upload bot could then use that for uploading.
Of course, any user of the tool then would have to give this information to my tools, without knowing what else I do with that (I won't, of course, but short of the toolserver admins, noone can actually confirm that).
OTOH, as long as I keep it optional, would such a mechanism be
acceptable?
Is there any other toolserver tool that requests or uses user's
passwords?
I really dislike the idea. I don't think the convenience outweighs the possible security problems and I don't think we should keep putting bricks onto temporary solutions: obviously the function should be performed by MediaWiki. But then, we already rely on CheckUsage, so... sigh...
Well, your password is transmitted unencrypted to through the internet every time you log into a Wikimedia project. So, there's already risk. The *additional* risk is in trusting me. That's everyone's decision.
Another possibility is that we ask for upload-from-url to be enabled on Commons. (I can't remember if it is an actual feature yet or just a mooted one.) Then CommonsHelper could be more or less single click, right?
Yes, it is an actual feature which has to be turned on by the-one-whose-name-we-must-not-speak, or one of his minions ;-)
No, it won't fix that problem; upload-from-url only offers a different source (web instead of your own PC); you'll still have to give a username and password to the script, one way or another.
It has been suggested that allowing upload-from-url will bring torrents of copyvios. Undoubtedly it will, but we get torrents of copyvios already. Upload-from-url should allow us to track the true origin of images much more easily (assuming that info is made available!). At the moment it is a lot of guesswork and hunches.
Yes, there might be more copyvios than now, but they'll be easier to hunt down and exterminate, and if some "violators" use this instead of the current upload, it would reduce overall stress on the admins, IMHO.
Magnus _______________________________________________ Commons-l mailing list Commons-l@wikimedia.org http://mail.wikipedia.org/mailman/listinfo/commons-l