Last modified: 2009-01-05 17:16:58 UTC
Gentlemen, on pages like http://commons.wikimedia.org/wiki/Category:PD_US_Military there is a big problem. If any file's name starts with a digit, no thumbnail is produced. Instead the whole giant image is pulled into the user's browser. This is a big waste of bytes. The software (please reassign bug if necessary) somehow thinks that the image must already bear a NNpx number-pixel name, and must already be a thumbnail, but this is not the case. Of course this waste is hidden through many, many, many layers of CSS. And there is also HTML "height=" and "width=". But if the user is using a text browser like emacs-w3m, the giant images "blow up in his face" to the true reality of the situation: monster images amongst normal thumbnails. The images that sail through without being thumbnailed with the above example page are: $ GET http://commons.wikimedia.org/wiki/Category:PD_US_Military | perl -wlne 'm@src="http://upload.wikimedia.org/wikipedia/commons/./../(\d[^"]+)@&&print $1' 030910-N-0743B-003-patrol.gif 091506bhigh.gif 091506high.gif 1-501pir.gif 12-inch-railway-mortars.gif 28-1433a_Mauthausen.gif 777px-Marlborough27s_march_to_the_Danube_1704.gif 81mm-mortar-massa-italy.gif
Added better test case to URL, improved summary. CCing Aaron, he knows more about image scaling than I do.
*** This bug has been marked as a duplicate of bug 16451 ***