This is more of a pet peeve than a question since I presume there are no alternatives to the situation.
If you want to access POST'd data via mod_perl, you need to use APR::Table, since that is the object Apache2::RequestRec::headers_in returns. This is required to get the "Content-Length" field, needed by Apache2::RequestIO to read() the data.
But APR::Table adds 45mb to virtual memory size -- 10 times the memory required by Apache2::Const, RequestRec, RequestUtil, and RequestIO combined. Apache just receives this stuff as a string, obviously, and I've been looking thru the Apache C includes: it does not seem to me that getting the POST data afterward would require so much complication. And a CGI process will receive POST data directly on STDIN.
Am I the only one that sees a serious design flaw here? I'd rather use POST for tidyness, but it is not worth the ridiculous APR::Table overhead...I guess I could just provide read() with a massive number, but that seems a very poor practice.
In reply to mod_perl and APR::Table: why the madness? by halfcountplus
| For: | Use: | ||
| & | & | ||
| < | < | ||
| > | > | ||
| [ | [ | ||
| ] | ] |