On Friday, 1 March 2019 09:01:26 CET you wrote: > Maybe it helps: on the other side other watch files for the same > download site are working: > > * roaraudio > * muroar > * muroard > * roarplaylistd
Uh ? http://software.keep-cool.org/dl/ shows only "animals" and "ckport" Roar audio home page points to http://roaraudio.keep-cool.org/downloads.html which sends back correct headers: $ mojo get -M HEAD -v http://roaraudio.keep-cool.org/downloads.html HEAD http://roaraudio.keep-cool.org/downloads.html HTTP/1.1 Accept-Encoding: gzip Host: roaraudio.keep-cool.org Content-Length: 0 User-Agent: Mojolicious (Perl) HTTP/1.1 200 Ok Content-Length: 41156 Last-Modified: Wed, 30 Jan 2019 19:58:28 GMT Server: mini_httpd/1.19 19dec2003 Content-Type: text/html; charset=iso-8859-15 Date: Fri, 01 Mar 2019 09:13:25 GMT In any case, HTTP::Message decoded_content doc mentions: « If the "Content- Encoding" or "charset" of the message is unknown this method will fail by returning "undef".» Since «%c» charset is unknown, I'd say that HTTP::Response behavior is correct. Given that uscan is only looking for the content of <a> tags, I'd suggest uscan to work on response's raw content instead of decoded content. I.e. call $reponse->content instead of $response->decoded_content. Thoughts ?
