root@maths.qmc.ac.uk (A system manager) (10/29/88)
Unfortunately I have not been able to clarify the problem further at this end (Perhaps I should say that we do not have the sort of Unix/C/mail expertise one might find in a CS Department, so fixing this sort of problem is quite difficult for us - hence the reason for forwarding this to others who may know the answers). However, here are a few reactions... |>It seems quite clear that when your mail gets |>to us it bears a hostname containing a blank (i.e. ASCII 20 hex, SPACE). |I agree that you interpret the path as.. |From ai.leeds.ac.uk!"dcs.leeds.ac.uk "!dcs.leeds.ac.uk!ajcole My comment was based on the fact that the error is in the call to sendmail rather than the output from sendmail. I realise sendmail calls itself and the error may be in the calling of the daughter process... |>Our mail program (sendmail, the Berkeley Unix mailer) reads this space |>as a word separator on its command line and thus finds the remaining |>arguments |>do not make sense (since what should be argument n is now argument n+1). |I agree with this analysis. |>I am not sure if your headers are legal under Janet (i.e. whether Janet |>hostnames can contain a quoted string including blanks) but it seems |>in any case uneccessary. |I think that our headers are legal (I've just had a quick skim thro |the back of the Grey Book). Nobody else has ever complained before |(and certainly not found "the" space) and we are churning thro a |couple of hundred outgoing per day (from 4 or 5 ARs) - BUT I am quite |happy to be corrected.. We would be in no position to advise on this! (see above) |But YOU generate the space you mention - I agree agree that the |answer to your question "whether Janet hostnames can contain a |quoted string including blanks" is no but the question is what |(in our original message) caused your relay(?) or you to generate |the "dcs.leeds.ac.uk " We do not knowingly do so, but I have not been able to figure out where the blank first occurs (anyone know how to track that?). Your hypothesis about 'Via:' fields (see below) is consistent with the behaviour of the subsequent test mesages. However feeding sendmail with such a test using /usr/lib/sendmail -t -v did not produce the same error. |> So the simplest cure would be for leeds to make |>sure its headers contain no blanks (I am reasonably convinced the blank is |>generated at leeds as we have seen nothing like this on any other mail). |See above - as far as I can see the "s and blank are generated at |QMC - the question is where from? |>Otherwise we have to get somebody to re-write the mailer for BSD to allow |>quoted strings with blanks in the sender argment... | ????? See comment above about your being unique in this respect. But as far as I know we are using standard software and no special fixes (We believe the mailer is standard BSD and we use the rmail that comes with the UK sendmail configuration package.) So how can we be unique...? |Being serious for a moment, the only part of our header (here I am talking |for Leeds.dcs not Leeds.ai) which differs between AI & DCS is the Via: |field - here there really is a problem (ugliness) unless one is prepared |to have "." in ether host names (or patch sendmail by hand) one gets nasties |like Via: dcs.leeds.ac.uk (csvax1.ARPA); Mon, 24 Oct 88 14:59:02 GMT |I think your relay is picking the space from here.. Well of course I'd be pleased to blame it on cs.qmc....! But that may well not be the case... |Via: dcs.leeds.ac.uk (csvax1.ARPA); Mon, 24 Oct 88 14:59:02 GMT | ^ ????? |but you are UNIQUE in this respect - at least in my experience. This seems to be unique at both ends! It is perhaps because most places use Received fields rather than Via fields, which would account for us never having seen it before if the problem is here Malcolm MacCallum as postmaster