From: Mark M. <Mar...@ij...> - 2005-04-29 15:04:51
|
Ian, > ... how feasible > would it be for amavisd-new to join a header continuation line > consisting entirely of whitespace to the end of preceding line > (i.e. just delete the CRLF between the two lines)? > > This should not affect subsequent interpretation of the message by > another MTA or MUA as it merely constitutes a partial unfolding of the > header lines. Here is a patch to implement this: http://www.ijs.si/software/amavisd/unfold-ws.patch It is applicable to 2.3.0 and 2.3.1-pre1. (and with some fuzz to earlier 2.x versions) Michael, > I still think it would be nice to at least have an option to turn on an > Outlook/Exchange hack. I understand that what's being done by > Outlook/Exchange isn't RFC compliant but lots of applications have > optional workarounds to accomodate deficiencies in external > applications. ... > Would be great if amavisd could do the same. Even better, call it > something completely obvious like "MSOutlookExchangeBreaksRFCHack" to > bring attention to it. Then there could be code along the lines of: > > if MSOutlookExchangeBreaksRFCHack = True > if HeaderLine contains <CR><CR><LF> > then <CR><CR><LF> -> <CR><LF><CR><LF> The bare CR or LF were allowed by rf822, and rfc2822 places them in the obsolete syntax (specifically: obs-utext). Treating bare CR as a line terminator may fix your case, but break others. It may perhaps not do much damage if done in the middle of a header, but allowing it to constitute a boundary between the header and the body would be a risky bussines and plain wrong - against both the rfc822 and rfc2822. Mark |