danw [Wed, 29 Jan 1997 23:09:37 +0000 (23:09 +0000)]
use gnu readline if the right variables are passed from config.Imakefile
throw some gratuitous #ifdefs around prototypes to make both POSIX and
non-POSIX systems happy with them.
insert a space between `=*' to make ultrix cc not give a warning.
danw [Wed, 29 Jan 1997 23:06:09 +0000 (23:06 +0000)]
${CRYPT} in Imakefile
same sanity fixes from the sipb-athena port.
several s/#endif lint/#endif/ to make irix cc happy
rename SignalHandler to Signal_Handler to avoid namespace collision in Linux
POSIX and other fixes in menu.c. Kludge for NetBSD since there's no
portable fix. Removed Password_input since it's not used and it was
hopelessly confused about POSIX vs SYSV
danw [Wed, 29 Jan 1997 23:03:24 +0000 (23:03 +0000)]
Get X/Motif library locations from config.Imakefile
#include <stdlib.h> instead of extern char *malloc();
POSIXification in tty.c from the sipb-athena port
danw [Mon, 20 Jan 1997 18:16:45 +0000 (18:16 +0000)]
Oracle changes
(sqlplus prefers input files to have names ending in `.sql')
newdb, newmoira, etc were out of date: New versions should be done
before the public release.
danw [Mon, 20 Jan 1997 18:14:06 +0000 (18:14 +0000)]
Oracle and Solaris/POSIX changes
Source tree should only have Imakefiles, not Makefiles.
Added ;s to awk scripts so Emacs will indent them properly.
Removed ancient files
danw [Sun, 17 Nov 1996 23:14:58 +0000 (23:14 +0000)]
Three bugfixes and a feature
- blank lines no longer give: Unknown request ""
- ^C just cancels the current line instead of killing the program
- \\ is now valid in quoted strings (along with \nnn and \")
- If you define USE_READLINE and tell it where to find includes
and libraries, it will build in support for gnu readline. This
isn't enabled by default, but I'm going to rebuild them with it,
and it will be easily configurable in the future autoconfiscated
version of the source tree.
danw [Tue, 29 Oct 1996 21:56:05 +0000 (21:56 +0000)]
new, more syntactically useful mrtest
allows "foo \"bar\" baz" or "\106\157\157" as arguments (useful for
running queries from moira logs)
It no longer uses the `ss' library, and it doesn't implement some of
the commands that no-one remembers what they do any more anyway. (test
and _compare_query).
danw [Sat, 12 Oct 1996 22:39:13 +0000 (22:39 +0000)]
Append numbers to MUL entries in filsys.db for use with new version of
attach on machines with new versions of BIND that don't always return
the entries in the same order.
If a printer doesn't have a quotaserver, output the printcap.db line
with no `rq' entry instead of the dubious `rq=(null)'.
danw [Thu, 10 Oct 1996 20:28:44 +0000 (20:28 +0000)]
Warn/help users who probably got the arguments in the wrong order:
danw@technomage:blanche> ./blanche danw -a some-list
WARNING: "LIST:some-list" was just added to list "danw".
If you meant to add yourself to the list "some-list", type:
blanche danw -d some-list (to undo this)
blanche some-list -a danw (to add yourself to that list)
danw@technomage:blanche> ./blanche danw -d some-other-list
./blanche: Unable to find member some-other-list to delete from danw
(If you were trying to remove yourself from the list "some-other-list",
the correct command is "blanche some-other-list -d danw".)
The warning message is given when the user tries to add a list to the
list with his/her username, and the help message is given when the
user tries to delete something from the list with his/her username and
it fails.
Additional, unrelated change: make the usage note document
`-db/-database' instead of `-S/-server' (to match the moira client).
The old options are still supported, just not documented.
* get_hostalias - If the machine name has no wildcards, look up
its mach_id and then do the query indexed by that instead of
by the machine name. This takes less than a second, vs. up to a
minute or so the other way. (This speeds up the `show machine
info' command in moira, which is used a lot by the RCCs this time
of year.)
* get_user_account_by_login - the sipb-athena moira client was
built from an older source tree that didn't have the patch
to never call gual with a string longer than 8 characters.
I hacked access_login to notice this and return an error.
(Because otherwise, it takes the server several minutes to
return `permission denied' when a user tries to use the `create
mailing list' command.)
Neither of these will be propagated to the Oracle server since it
doesn't have a problem with either of those queries.
danw [Wed, 21 Aug 1996 03:30:13 +0000 (03:30 +0000)]
applied the patch from [887] in moira, to fix addresses that appeared
without an `@mit.edu'. (The version of pobox.gen on moira2 already has
this patch.)
danw [Mon, 12 Aug 1996 20:59:35 +0000 (20:59 +0000)]
Make sq routines check the result of malloc. (Mostly for dbck.)
sq_save_data, etc now return 1 on success and 0 on failure. (This is a
backwards-compatible change: they were implicitly returning int values
already, but those values weren't being checked anywhere).
In the past year, added: /mit/info/public-mailing-lists maintained
nightly, made backup paths more consistent, compares backup file sizes
with numbers in a conf file to make sure backup really succeeded.
dkk [Wed, 6 Mar 1996 04:19:32 +0000 (04:19 +0000)]
This includes Ted's unchecked-in changes from two years ago, as well
as recent changes to support: default shell /bin/csh=>/bin/athena/tcsh,
default status 4=>0, and a few little changes.
jweiss [Tue, 17 Oct 1995 22:58:27 +0000 (22:58 +0000)]
WIFEXITED(waitb) is true if the script exits even with exit 0.
changed so we only return an error if something other than an exit 0
occurs. Note that the error returned in the case of death doe to a signal
will be a bogus error.