11/14/79 If .mh_profile contains an entry like "show: -pr -a -b -c" and you call "show" with the explicit switch -nopr, then you have those dangling -a -b -c switches that were intended for pr. (Bound the switch list for pr, etc., with a semi-colon?) 2/1/80 "file +junk" makes junk a folder, while "file +.junk" makes ".junk" a subfolder. "file" should test for pathname beginning with "./" instead of just '.' 9/15/80 When repl gets an adrparse error, it should return to "What now?" instead of aborting. (But see objections in /r/phyl/Mail/MH/16) The same goes for comp, dist, forw 9/29/80 When send is invoked with no args, and the draft already exists, options to "use draft?" should include "list" along with yes/no 10/21/80 Tabs in address lists aren't recognized. 12/23/80 Fix and install "dist" 1/14/81 inc -nochangecur displays that the first-inc'ed message is the current message, and it's not. It should also do a scan from cur, I think. (I fixed the 1st bug, but not the scan from cur. /phyl 7/9/81) 5/28/81 'news' will break when there are multiple news folders in the 'To:' field(s)! 7/7/81 phyl Re:Log:"3/23/81 jdg/day: changed deliver.c to check for effective uid and gid while it is preventing forgery of mail. Needed for uucp, when uucico is invoked by a real person instead of uucp itself." The real fix should be to check for real uid==0 (srvrftp) or real gid==daemon (any uu program) or effective gid==daemon (human running uucico, and uucico should be SGID daemon). 2/16/81 /usr/tmp/mh files are readable by the world. So are /usr/spool/uucp files. 2/21/81 A message should not have to exist for you to be able to use its number in a range, as in 'scan cur-last' when cur does not exist. 2/28/81 "This RAND-UNIX stuff (adding and removing them from message headers on the vax) is a crock. Reply strips 'RAND-UNIX' off of addresses, but not 'rand-unix'. If I type it up myself, deliver rejects either of them." --guyton 4/4/81 "The delivery program should not leave headers that the mailer doesn't understand. Replying to the following header was painful: " --jim (Message inbox:1) From: graphics!root Date: 4 Apr 1981 at 2138-PST To: Vax!mh-discussion at RAND-UNIX Cc: greep at SU-ISL Subject: Mail to terminals From: greep at SU-ISL ...AND LOTS OF OTHER MESSAGE HEADER OBJECTIONS 4/7/81 scan and inc should check ALIASES and recognize UUCP localhost!me when deciding whether or not the message is from me. 4/16/81 Deliver should run one uux per site instead of one per addressee. 6/2/81 "rmf" should automatically remove .e* files from the named folder. 7/81 deliver fails to recognize "at" when it is preceded by more than one blank. repl tries to parse a Sender [or From] field, and a To field, without first testing for their existence. file protections on rmail/uucico are inconsistent with deliver's internal user/group tests. deliver fails to recognize that localhost!user is indeed user. (Should it?) temporary mail files stay around when the "no such local user" error occurs--because of a spurious flag test. deliver should parenthesize (user at host) as well as (host!user) when constructing the uux command. rmail should direct stderr to a file before invoking deliver. rmail should check for unknown local user and send return mail to the originator. (Can't fix unless no intervening Arpanet addresses.) rmail should add a Date: component if there is none. mail headers should be reordered. 7/23/81 "mail" should be in /bin, not in /usr/randbin. (?) 8/17/81 "rmf" should verify that the requested folder is not the current directory. 8/20/81 /usr/tmp/mh and /usr/MailAliases should be defined in strings.a 9/1/81 repl should remove leading blanks from date and subject fields before putting them in the new in-reply-to and subject fields. (Greep) 9/9/81 Adrparse should recognize the full RFC733 syntax; viz, colon as part of the address list. 10/2/81 Subject: news bug From: phyl at RAND-UNIX The table of news topics overflowed. news.c counts all non-dot files in the directory /usr/news to determine the number of news topics. But mail sent to an UNKNOWN news topic remains in the top level news directory, erroneously getting counted into the total. Either .mh-receive should stow undeliverable news into another directory, or news.c should count only directory entries. 10/6/81 mhl should know about termcap 10/14/81 Subject: Mail to "everyone" From: obrien at RAND-UNIX An insoluble problem: sending mail to "everyone" (a risky business at best) generates mail to every line in /etc/passwd, including those like "uucp", "unix32v", and such which aren't real people. An exception list is no help because it gets out of date. Only cure I know of is to go through /usr/spool/mail every now and again and get rid of phony files. 11/10/81 Subject: HOSTNUM not necessary From: greep at SU-DSN Mh deliver should really not need HOSTNUM (this is in /usr/include/whoami.h) - it can get the number by searching for HOSTNAME in the host table. 11/18/81 SUBJECT: folder with a -fast flag doesn't change current folder From: norm at RAND-UNIX 11/81 Message lists should not be reordered. (Norm, Phyl) 12/17/81 The alias function doesn't work when a single message contains multiple instances of the address form "foo.*"--ergo, a message cannot be sent to more than one news group. Note that /usr/news/.mh_receive won't recognize multiple news groups in a single message either! 12/19/81 If a message is addressed to more than one local alias-group, people who are in more than one of the groups receive multiple copies of the message. (I think! --phyl--) 1/5/82 If you send a message to a bad uucp host name, it a) gives you an error message, b) queues the mail anyway, and c) deletes the draft because the mail was "correctly sent". Ha! It should a) give you an error message and b) punt. --Michael W.-- 1/19/82 MH argument parsing should be upgraded to allow folder arguments to have a space between the + and the folder name so you can use shell matching syntax. Example: % folder + /usr/news/net-g* % cd !$ --dave 3/11/82 deliver should clean up before exiting.