On 2010-04-21, Alan McKinnon <alan.mckin...@gmail.com> wrote: > On Tuesday 20 April 2010 15:53:01 Harry Putnam wrote: >> I think you all are missing something... sendmail is better documented >> than any of the other pretenders. > > One has to understand what the various MTAs out there were built to do, and > what their "feature list" is: > > sendmail comes from ancient days. It was written to be able to route almost > any kind of mail using almost any kind of addressing scheme to and from > almost > any kind of network.
Very true. And since nobody (that I know of) needs that capability any longer, asking modern Linux users to continue to pay for that capability everytime they try to tweak the MTA configuration seems a tad silly. > So it is quite happy receiving SMTP mail from the internet and > routing it to a FidoNet address. To do this, it has to reread it's > routing table with every message, therefore .cf was designed to be > machine efficient but still use only ASCII characters. Which led to > m4 being developed Sendmail didn't lead to m4 being developed. m4 was developed by K&R in the mid 70's. Sendmail didn't happen until the early 80's. According to Wikipedia, sendmail first shipped with BSD 4.1c in 1983. Unless in this context, m4 doesn't refer to the m4 macro processor and associated language? I always thought that the m4 used to encrypt sendmail configurations was the standard Unix m4 that was developed for Ratfor in the 70's. Wikipedia seems to confirm that, saying that "The implementation of Rational Fortran used m4 as its macro engine from the beginning", but Wikipedia also says that m4 was developed in 77 and Ratfor in 74. Both were developed by K&R, so I suppose it could be that m4 was used by Ratfor for a couple years before m4 went public as a seperate program. > Even a cursory glance at sendmail shows that it was designed in a > time with a different mindset and different needs to what we do these > days. Sendmail will never escape this legacy because it is what it is > and that is it's purpose. > > It's not as bad as buggy whips, but the same principle is at work. The UHH chapter on sendmail has some great examples of sendmail address parsing/transformation run amok. -- Grant Edwards grant.b.edwards Yow! My polyvinyl cowboy at wallet was made in Hong gmail.com Kong by Montgomery Clift!