Right now, any registered user can upload a file with the same name as an already existing one, thereby either unintentionally or deliberately overwriting and removing the original file. This will be recorded on the upload logs page, but not in the history of the page which contains the image. As a result, anybody who uploads images needs to keep local safety copies and needs to check from time to time to make sure that the file on the server is still the intended one.
I propose that we completely forbid the overwriting of existing uploaded files. If you have a new version of your image, give it a different name and change the link in the article. That way, the change will be recorded in the article's history, and the previous version of the image will still be accessible.
Sysops would of course still be able to remove any file.
Axel
On 5/19/02 3:25 PM, "Axel Boldt" axel@uni-paderborn.de wrote:
Right now, any registered user can upload a file with the same name as an already existing one, thereby either unintentionally or deliberately overwriting and removing the original file. This will be recorded on the upload logs page, but not in the history of the page which contains the image. As a result, anybody who uploads images needs to keep local safety copies and needs to check from time to time to make sure that the file on the server is still the intended one.
I propose that we completely forbid the overwriting of existing uploaded files. If you have a new version of your image, give it a different name and change the link in the article. That way, the change will be recorded in the article's history, and the previous version of the image will still be accessible.
Sysops would of course still be able to remove any file.
But that's annoying if you want to update a file with a corrected version. There should be a better way.
At 03:38 PM 5/19/02 -0400, you wrote:
On 5/19/02 3:25 PM, "Axel Boldt" axel@uni-paderborn.de wrote:
I propose that we completely forbid the overwriting of existing uploaded files.
But that's annoying if you want to update a file with a corrected version. There should be a better way.
It would be nice if uploaded files had a "history" like articles do. That way a vandalism like overwriting with junk could be easily undone.
Also, while I'm thinking about it, a "trash bin" where deleted files and articles go for a while before being truly and permanently deleted would be a nice safety feature; someone was just recently asking about possible ways to recover an article he'd deleted, and I myself am often reduced to a quivvering mass of indecision and fear when I'm about to click on "delete this page" even if I've spent considerable effort convincing myself it's the right thing to do.
Oh, and an "articles that link here" feature for the files in uploads would be great, too! Would make it easier to spot the files which are actually used by articles and the ones put there by jerks just using Wikipedia for free hosting space.
I think I'm done wishing for now. :)
-- "Let there be light." - Last words of Bomb #20, "Dark Star"
On dim, 2002-05-19 at 12:50, Bryan Derksen wrote:
At 03:38 PM 5/19/02 -0400, you wrote:
On 5/19/02 3:25 PM, "Axel Boldt" axel@uni-paderborn.de wrote:
I propose that we completely forbid the overwriting of existing uploaded files.
But that's annoying if you want to update a file with a corrected version. There should be a better way.
It would be nice if uploaded files had a "history" like articles do. That way a vandalism like overwriting with junk could be easily undone.
Hmm...
Also, while I'm thinking about it, a "trash bin" where deleted files and articles go for a while before being truly and permanently deleted would be a nice safety feature;
That would be pretty easy, I think. Just create a separate directory (with web access disallowed), move 'deleted' files over it; maybe have a cron job clear it out every now and then.
Oh, and an "articles that link here" feature for the files in uploads would be great, too! Would make it easier to spot the files which are actually used by articles and the ones put there by jerks just using Wikipedia for free hosting space.
This would be easier if we add a table to list external links just like we currently have for internal links to existing and nonexisting articles. When saving an article, we could list all the web links, images, and perhaps inter-wiki links, dump them to this table; then, it's a simple database query to check against the URL and grab any article -- on _this_ wiki -- that links to a given file.
(If there's any reason not to do it this way, let's discuss further on wikitech-l; however it's implemented, it's a needed feature.)
This will not solve the problem of tracking files that are used on _other_ wikipedia sites; a lot of images are shared between articles on the different languages 'pedias, and a number of specifically non-English files (labelled maps and diagrams, for instance) are hosted on the English or Meta 'pedias because the others haven't been converted yet... though theoretically this could be tracked in the future by checking the link tables for each language's database.
-- brion vibber (brion @ pobox.com)
wikipedia-l@lists.wikimedia.org