On 16.11.2007 17:08 Michael Peters wrote: > Is this a mod_perl problem? You didn't mention if it occurred in plain CGI or not.
It would be difficult to port the application to plain CGI but you are right, I could have built an independent test script. But because I could swear I did UTF-8 in plain CGI in the past I was quite sure it was a mod_perl issue. >> The script with the extended characters is written in UTF8 but the >> resulting page has the extended characters in latin1. >> Background: This is a runmode of a CGI::Application but it shouldn't >> make a difference because I don't use the CGI::Application magic >> (sending headers and output of the result) in this runmode. > > Why aren't you using C::A's help in this case? I've done UTF-8 in C::A before. Because I didn't use the usual C::A's content handling (via return) and therefor expected an "ask somewhere else". It is really difficult to guess the correct forum as long as you don't know the real source of the problem... > The key is to tell CGI.pm that you're using UTF-8 (but get v3.30 from RT since > 2.29 has UTF-8 issues). Cannot find it but since it is working now I think I can wait for the release of v3.30. >> Any ideas what may be going wrong here and why the output is converted >> back to latin1? > > CGI.pm uses STDOUT to send the contents. So we need > > binmode(STDOUT, ':utf8'); that did the trick, many thanks! - Michael