On Thu, 12 Feb 2004, Lloyd Zusman wrote: > Well, from what I know about the structure of Courier, it would take > a lot of refactoring to allow the modification of messages during > a global filter. That's because the message file that we see during > this step is a temporary file. The "real" message file has not yet > been created, as far as I know. > > You can see the sequence of events that take place during message > processing here: http://www.courier-mta.org/queue.html
I don't think that's quite accurate. My understanding of that same document is that we are operating on the same exact message that ends up being processed by courierd, but it's *current name* is such that courerd ignores it until submit renames it. However, it's not like when filtering is done, submit creates *another* copy of the message, it's just renamed. > And as for this patch itself, remember that it consists solely of > putting a unique "id" field into the "Received" header. This is a minor > change, and it mirrors what some other MTA's already do. Even if we > don't end up using this to facilitate the methodology that I outlined in > my Modest Proposal, it still is a useful feature in and of itself. To me, it's just a hack. Sorry, but that's how I feel. > > I've seen people running filters twice to do rejection and modification > > separately - maybe before embarcing on this you might want to do some > > benchmarks? > > I fully intend to perform some benchmarks to _compare_ the > rejection/re-insertion idea with my proposed methodology. I can only do > this once I have written the code outlined in my proposal. This will > be completed in a few days, at which time I'll run the benchmarks and > post the results here ... as well as my patch and code. I think that's a great idea! However, my suggestion would be to use submit to your advantage here. The "submit" protocol is well documented somewhere (I read it for over an hour the other night) and while a little unwieldy IMO it also seems like the best way to do injection. I come from a qmail background, and it seems to me that 'submit' takes the place of qmail-inject and qmail-queue. It also seems like there might be some benefit to writing a qmail-inject-like wrapper for submit. -- Ensign Walnut approaches Dr. Crusher with caution... Jon Nelson <[EMAIL PROTECTED]> C and Python Code Gardener ------------------------------------------------------- SF.Net is sponsored by: Speed Start Your Linux Apps Now. Build and deploy apps & Web services for Linux with a free DVD software kit from IBM. Click Now! http://ads.osdn.com/?ad_id=1356&alloc_id=3438&op=click _______________________________________________ courier-users mailing list [EMAIL PROTECTED] Unsubscribe: https://lists.sourceforge.net/lists/listinfo/courier-users
