http://archive.aravot.am/ is encoded in ARMSCII-8 but is declared as x-user-defined, which makes the site fail to render correctly out-of-the-box in any browser but makes the site readable in Chrome and IE if the user has installed an intentionally mis-encoded font called Arial LatArm. To make the archives readable without special trickery, the site should convert the old files from ARMSCII-8 to UTF-8 and change the declaration from x-user-defined to UTF-8. Alternatively, once bug 213517 lands, the site could (font license permitting) add Arial LatArm as a @font-face font, which would make the site work visually cross-browser without users having to have the font installed. Conversion to UTF-8 would be a better solution, though, because unlike the font hack, it would make clipboard and search work.
TBH I can't tell if this bug still exists or not, but the one page contact info that could be tried. If anyone speaks Armenian. Yerevan, Republic of Armenia, 2 Arshakuniats, 15 floor tel. 568 968, 586 529, fax 528 752, E mail email@example.com
>> curl http://archive.aravot.am/ | grep charset >> <meta http-equiv="Content-Type" content="text/html; charset=x-user-defined"> It's at least still declaring x-user-defined.
INVALID :) http --print h GET http://archive.aravot.am/ HTTP/1.1 500 ( Connection refused ) Cache-Control: no-cache Connection: Keep-Alive Content-Length: 4288 Content-Type: text/html Pragma: no-cache Proxy-Connection: Keep-Alive Via: 1.1 NS