Graham Leggett wrote:
I'm neither the author of mod_headers nor can I see a bug in the current behaviour.Maik Mueller wrote:I don't consider the behaviour of mod_headers as a bug. The point is that the content of the environment variable SSL_CLIENT_CERT isn't formatted according to the rules for folding headers onto multiple lines.
The job of formatting the header onto multiple lines should be the job of mod_headers - the data should come out the other side exactly as it was passed in. Fixing this to work correctly is a far more elegant solution to the problem.
However, I admit that your proposal will improve the ease of use of environment variables in headers.
You have to do both in any case. The check itself causes the performance penalty.Building a correct header from a arbitrary environment variable's content isn't that easy (CRLF, character escaping,...) and can be a considerable performance penalty.
Not necessarily. All it takes is to add a space before each line sent out, which is not difficult to do.
As to character escaping, I am not 100% sure what character set restrictions exist on the content of headers.
And the inelegant thing is, that you waste the time even if you don't have to change anything.
In your case the characters are base64 characters, which are not a problem. RFC2616 should cover this though anyway.Correct, but you have to check the data before you know that.
By the way I'm not the one to defend the current implementation of mod_headers.
To my mind this part is very performance critical and I have tried to affect performance as little as possible - especially when not using my new feature.
Probably we should look for other opinions.
Regards,
Maik
