Because we haven't had the voters to approve a release subsequent to apr-iconv 1.2.1, and this code is significantly out-of-date, I'd like to propose we shutter the codebase.
Since we grabbed that particular starting point, GNU libiconv became the maintained reference standard for glibc. IBM's ICU is very, very heavyweight and does streaming. But to have a license-compatible implementation, I've been paying attention to the bsd family of OS's. The citrus project appears to be the right solution for Windows, and perhaps any unix with a still-broken iconv implementation (do any examples still exist?) I'm proceeding to fork for porting and introduce a cmake build structure for the project. I'll look at the other few porting efforts to see if there is a basis for some collaboration. Because this is 'just iconv', build exceptions we have in apr disappear. It might be worthwhile to support sideloading of citrus or libiconv distinct from the clib (this has historically been done in libiconv using an alt namespace, with the include header masking the change). So if you would like to see apr-iconv persevered, please speak up now, and please start bug-fixing because if we don't kill it, we need to maintain it. Cheers, Bill
