User talk:FlickreviewR/archive 1

From Wikimedia Commons, the free media repository
Jump to: navigation, search

Bugs[edit]

Please add bugs to the table. I use this table primarily for myself as a reminder, but everybody is free to add.
Syntax:
| [link to diff] || ~~~~ || (comment (optional)) ||
|-

Diff Reporter Comment (optional) Comment by developer
[1] Atamari 19:27, 18 November 2006 (UTC) Fixed -- Bryan (talk to me) 20:58, 18 November 2006 (UTC)
[2] Bryan (talk to me) 20:28, 18 November 2006 (UTC) Fixing links where username does not match real name Fixed, but unfortunately requires all images to be rescanned. -- Bryan (talk to me) 21:50, 18 November 2006 (UTC)
[3] [4] [5] -- Bryan (talk to me) 19:43, 19 November 2006 (UTC) Link not found... again... Confused by my own vague naming ;) Commenting code should really help -- Bryan (talk to me) 22:40, 25 November 2006 (UTC)
[6] [7] -- Bryan (talk to me) 21:01, 19 November 2006 (UTC) Exif errors
[8] Solipsist 12:25, 21 November 2006 (UTC) Bot clearly isn't checking against the right Flickr image nor author. Nope, that is because it has no exif information. See the request page why it doesn't check. -- Bryan (talk to me) 20:14, 21 November 2006 (UTC)
The tag the bot is adding still looks bogus to me. AFIK, this image has nothing to do with hisglassworks on Flickr. -- Solipsist 00:34, 22 November 2006 (UTC)
Ok, now I see what you mean. I'll try to fix it before the next run. Luckily the bot also does an EXIF comparison, so it is at least not wrongly tagged passed. Thank you for you comment, and I will fix it before the next run. -- Bryan (talk to me) 15:54, 22 November 2006 (UTC)
Fixed -- Bryan (talk to me) 22:07, 25 November 2006 (UTC)
[9] -- Bryan (talk to me) 12:38, 26 November 2006 (UTC) The self tags Fixed, as well as the lang-cc tags. -- Bryan (talk to me) 21:09, 3 December 2006 (UTC)
[10] Lugusto҉ 02:04, 29 November 2006 (UTC) The bot have marked this image as not found, but this image exists in the source URL. Lugusto҉ 02:04, 29 November 2006 (UTC) The bot marked this image as "NO_EXIF", which means that it cannot compare the image on flickr and the image on commons. It's a feature, not a bug ;). This is done to prevent the bot from authorizing images that actually point to a flickr page that is not the same image. For those images without EXIF, I am however planning a bit by bit comparison, which I hope to be live in 2 weeks. -- Bryan (talk to me) 15:34, 29 November 2006 (UTC)
[11] -- Bryan (talk to me) 21:09, 3 December 2006 (UTC) Link not found
[12] -- Bryan (talk to me) 22:00, 4 December 2006 (UTC) Matching did not work correctly
Para 07:06, 23 December 2006 (UTC) Most of the images the bot has added to Category:Flickr images not found are still on Flickr. Seems to be fixed, the problem was a new algorithm that was meant have a higher hitrate for detecting flickr links. -- Bryan (talk to me) 11:03, 23 December 2006 (UTC)
[13] --:Bdk: 00:59, 3 January 2007 (UTC) needless upload (third version of this image), "original size" image was already there (see second version). seems as if only the first upload to Commons is recognized. (noted several more cases like this after my first edit here, see [14] for another example) Should be fixed by know. Test run starting. -- Bryan (talk to me) 13:54, 3 January 2007 (UTC)
[15] — coelacan — 22:30, 17 January 2007 (UTC) Marked this image as PASSED_CHANGED even though the license noted is the one the bot recorded (cc-by-2.0). Thank you, it didn't recognize the atribution argument on the CC tag. I'll fix it. -- Bryan (talk to me) 22:34, 17 January 2007 (UTC)
[16] Thinks image is gone (it is still there). -Susanlesch 07:51, 1 April 2007 (UTC) Thanks for reporting. Has been fixed. -- Bryan (talk to me) 14:01, 1 April 2007 (UTC)

General comments[edit]

md5 hash[edit]

Wouldn't comparing MD5 checksums make more sense? Lots of images don't have EXIF data. —Chowbok 01:24, 12 December 2006 (UTC)

Yes it would. However, since MW does not supply any md5 anywere, and neither does flickr, I have to generate them myself, which means that they have to be downloaden. Doing it this way, it will take about 45 seconds per image, which is to slow. I have however implemented this, and I will be rerunning the bot on images with no EXIF information. This function is still in test phase. Three images have been tested so far. [17] -- Bryan (talk to me) 11:49, 12 December 2006 (UTC)

NO_EXIF 5[edit]

Not sure if this is a bug or where it fits above, but it seems that the bot is unable to check out most of the images I have fetched from Flickr. A few have passed, but looking at these and comparing the Flickr pages they link to, I fail to see any difference from the ones that didn't pass. Celsius 12:12, 30 December 2006 (UTC)

For instance: What is the difference between Image:Uppsala Cathedral - Grave of Bengt Oxenstierna.jpg (with indeterminable copyright status) and Image:Uppsala Cathedral - Epitaph of Linnaeus.jpg (which was confirmed). Both come from the same set of photos by the same Flickr user, have the same license, and were uploaded by me on Commons within a few hours from one another. Celsius 12:19, 30 December 2006 (UTC)

The first one has no EXIF data, the second one does. I see that they both have EXIF data on Flickr, so somehow it got lost on that first one. Did you edit the image in any way? Some programs aren't good about saving that. You may want to re-upload the first image and ensure the data is retained. —Chowbok 16:53, 30 December 2006 (UTC)
I don't remember if I edited the image, but I sometimes do. I guess this could be the explanation. I'll avoid doing that in the future. Celsius 21:13, 30 December 2006 (UTC)

Yes, it is true and by design that images without exif information are not checked. I have however implemented a phase 2 reviewer, this will fetch the whole image and compare them. This demands a lot of both bandwith and cpu power, so I plan to do that once I have finished the normal tagging. Thanks for your remarks, -- Bryan (talk to me) 12:27, 30 December 2006 (UTC) (Moved here from my talkpage, as it may be useful to someone else with the same question. Celsius 21:13, 30 December 2006 (UTC))

That said, I would recommend that people try to retain EXIF data even if it is no longer necessary for the bot, as it contains a lot of useful information in general. —Chowbok 01:42, 31 December 2006 (UTC)
Indeed -- Bryan (talk to me) 15:13, 1 January 2007 (UTC)

Sourcing for non-Flickr images[edit]

Would it be possible to use a modified form of FlickreviewR to search for non-Flickr images sourced to webpages? Jkelly 05:22, 28 January 2007 (UTC)

No. FlickreviewR uses the Flickr api to check the license of the image. So this generally can not be used for random websites. -- Bryan (talk to me) 15:50, 28 January 2007 (UTC)
Thanks. Jkelly 22:13, 28 January 2007 (UTC)

Flickr validity?[edit]

Obviously the bot can only check for what Flickr's page says, not for whether the Flickr page itself is correct or not. On Image:Dakota Fanning.jpg, I updated the image description page because I doubt the voracity of the ability of the Flickr uploader. My information on this image was removed by the uploader MatthewFenton (talk · contribs), but still seems pertinent.

More than this one questionable image, I'm concerned with the legitimateness of images which Flickr uploaders may not have the rights to, but claims Creative-Commons copyrights on anyways. Do we have any recourse for this possibility of fraud? — pd_THOR | =/\= | 17:47, 1 February 2007 (UTC)

Flickr users generally don't have a clue about copyrights. This image is thus clearly a copyright violation by the Flickr user and will be deleted. -- Bryan (talk to me) 17:50, 1 February 2007 (UTC)

How often?[edit]

It might be a good idea to mention on the page how often the bot checks the Flickrreview category. Daily? Weekly? Hourly? Ekantik

Less often than it used to be. There are still database problems which really need to be fixed now. I am currently having an account on the Toolserver, so I plan to have it run continiously, but it will take me some time to set it up, as I also want to convert the sqlite database to mysql. But currently the COM:POTY/2006 has a higher priority for me, and I really need to finish stuff for that before 1 March. -- Bryan (talk to me) 09:41, 24 February 2007 (UTC)
I see, hope it all goes well. In the meantime, is it possible for you to perform a human review on this image? For the record I was instrumental in getting this image properly licensed (although someone else uploaded it obviously) so I can confirm that the image has been licensed appropriately at my instigation. I would request a human review asap because the image is in use at w:Shilpa Shetty and had just been deleted because of potential copyvio, so I would like to hurry up the process if possible. Thanks, Ekantik 03:56, 5 March 2007 (UTC)