gnu@sun.UUCP (08/30/84)
Ahem. A moment of reality, please. All this software was written under time pressure and is still evolving. Precious little of it was originally written or designed to do what it is currently doing. Five years ago, few of these systems had even rudimentary capability to hook up to foreign networks -- many did not know how to pass data with another machine of their own kind! Being in the "leading edge" of mail technology, being in the biggest most advanced and interconnected networks in the world has its prices: things in thousands of hosts scatterred all over the world do not always, necessarily, perfectly, work right every time. I don't have time to pop into sendmail every time someone reports a bug; my sendmail bug folder has several hundred messages in it. But I regularly send and receive mail with people in Europe and on corporate networks, which I could not do two years ago. If you have a problem with somebody's headers, please, tell us about it. But one tortured scream, and maybe a rebuttal, should suffice. We can hopefully expound our theories of "the way to run the mail network" without getting our throats this hoarse. Personally I have written many programs that would coredump when given invalid input data, but I didn't write them as production programs, I wrote them as hacks. In time of need, hacks are pressed into service and fixed later. This is not a cause for alarm; the alternative to Unreliable mail service is No mail service, which is the default when you first apply power to your chips and can continue in force for quite a long time if desired. As soon as we find a reliable mail service we'll be sure to tell you, so we can talk to you with it.