MOO-cows Mailing List Archive

[Prev][Next][Index][Thread]

Re: ALPHA-TEST release of LambdaMOO version 1.8.0alpha6



Has ANYONE compiled this on WindowsNT or Windows95?

Kevin

> From:          Pavel Curtis <pavel@parc.xerox.com>
> To:            MOO-Cows-Announce@parc.xerox.com
> Subject:       ALPHA-TEST release of LambdaMOO version 1.8.0alpha6
> Date:          Tue, 16 Jan 1996 00:06:27 PST

> I have made the next (and perhaps the last) *alpha-test* release of version
> 1.8.0 of the LambdaMOO server in the usual place:
> 
> 	ftp://ftp.parc.xerox.com/pub/MOO/LambdaMOO-1.8.0alpha6.tar.Z
> 
> This is still an ALPHA TEST release of the server and, although it's getting
> more and more stable as time goes on, there are almost certainly some more
> server-crashing, database-losing bugs remaining, so please DO NOT USE THIS
> RELEASE IN A PRODUCTION SETTING!
> 
> Please continue to send your bug reports to the main MOO-Cows mailing list.
> I still expect to make at least one more test release before the final,
> production-approved one, but I'm now hoping that that will be a `beta' release
> (i.e., with no new major features to be added before final release).  We'll
> see, though.
> 
> Here's the ChangeLog.txt entry for this release:
> -------------------------------------------------------------------------------
> Version 1.8.0alpha6, 16 January 1996
> -- Replaced panic with raising E_TYPE in the case where `x[$]' is applied to a
>    non-string, non-list `x'.  (Thanks to Kipp the Kid for reporting this.)
> -- Fixed (yet another) bug in the stack-unwinding code; this one stuck the
>    return information for each of a built-in's second and subsequent verb calls
>    on successively higher stack frames, causing all manner of havoc, the least
>    of which was somewhat surprising results from the (correct!) callers()
>    built-in.  (Thanks once again to Brian Buchanan for reporting this bug; this
>    marks the *third* bug in stack-unwinding that Brian has uncovered!)
> -- Changed the rules for binary strings slightly so that (a) TAB is no longer
>    treated as a normal, printing character (i.e., it gets converted to `~011'
>    on input in binary mode), and (b) `~' is treated just like all of the
>    non-printing characters (i.e., it gets converted to `~176' on input in
>    binary mode and the sequence `~~' is illegal in binary strings).  Change (a)
>    removes one more source of TAB characters from the MOO (which can only be a
>    good thing), and change (b) both simplifies the specification and makes it
>    easier to compute the true length of a binary string (i.e., it's the length
>    of the string itself minus three times the number of tildes).  (Thanks to
>    James Deikun for these suggestions.)
> -- Fixed uninitialized-variable bug in the `program' command in emergency
>    wizard mode.  (Thanks to Kipp the Kid for reporting it.)
> -- Fixed potential memory smash bug if #100 (or #200 or #400 or #800 or ...)
>    was a recycled object in the DB file being loaded.  (Thanks to Dave Van
>    Buren for sending me a DB file that tickled the bug and to Purify for
>    pinpointing it.)
> -- Fixed bug that caused *all* settings of the .wizard property to be logged,
>    instead of just those that set it to a true value.  (Thanks again to Dave
>    Van Buren for reporting this.)
> -- Fixed bug accuracy of the top line number in tracebacks resulting from the
>    setting of a wizard bit.  (Thanks one last time to DVB for sending me a
>    traceback that contained such an error.)
> -- Added new built-in function `function_info([NAME])' which returns
>    descriptions of built-in functions available on the server.  If NAME is
>    provided, only the description of the function with that name is returned;
>    if NAME is omitted, a list of descriptions is returned, one for each
>    function available on the server.  Each description is a list of the
>    following form:
> 	{NAME, MIN_ARGS, MAX_ARGS, TYPES}
>    NAME is the name of the built-in function, MIN_ARGS is the minimum number of
>    arguments that must be provided to the function, MAX_ARGS is the maximum
>    number of arguments that can be provided to the function or -1 if there is
>    no maximum, and TYPES is a list of MAX_ARGS numbers (or MIN_ARGS if MAX_ARGS
>    is -1), each of which represents the type of argument required in the
>    corresponding position.  Each type number is as would be returned from the
>    typeof() built-in function except that -1 indicates that any type of value
>    is acceptable.  For example, here are several entries from the list:
> 	{"listdelete", 2, 2, {4, 0}}
> 	{"suspend", 0, 1, {0}}
> 	{"server_log", 1, 2, {2, -1}}
> 	{"tostr", 0, -1, {}}
>    `listdelete()' takes exactly 2 arguments, of which the first must be a list
>    (LIST == 4) and the second must be a number (NUM == 0).  `suspend()' has one
>    optional argument that, if provided, must be a number.  `server_log()' has
>    one required argument that must be a string (STR == 2) and one optional
>    argument that, if provided, may be of any type.  `tostr()' takes any number
>    of arguments and it can't be determined from this description which argument
>    types would be acceptable in which positions.
>       It should be noted that, in a feat of synchronicity, this new built-in
>    brings the total number supported in the official server release to exactly
>    100!  (Thanks to some-user-whose-message-I've-lost [feel free to mail me
>    again so that I can patch this thank-you note] for suggesting such a
>    built-in function.)
> -- Fixed the definitions of the built-in functions match(), rmatch(), index(),
>    rindex(), and strsub() to accept any type of value for their final,
>    `case-matters' flag argument.  Also fixed `verb_code()' to allow any type of
>    value for either of its final two flag arguments.
> -- Added `scattering assignment' expression, allowing the elements of a list to
>    be spread among multiple variables simultaneously; this could be used, for
>    example, to get at the arguments to a verb in a more convenient form than
>    the list `args'.
>       A scattering assignment expression looks like this:
> 	{ TARGETS } = EXPR
>    where TARGETS is a comma-separated list of places to store elements of the
>    list that results from evaluating EXPR.  A target has one of the following
>    forms:
> 	VARIABLE
> 	    This is the simplest target, just a simple variable; the list
> 	    element in the corresponding position is assigned to the variable.
> 	    I call this a `required' target, since the assignment is required
> 	    to put one of the list elements into the variable.
> 	? VARIABLE
> 	    I call this an `optional' target, since it doesn't always get
> 	    assigned an element.  If there are any list elements left over
> 	    after all of the required targets (and all of the other optionals
> 	    to the left of this one) have been accounted for, then this
> 	    variable is treated like a required one and the list element in the
> 	    corresponding position is assigned to the variable.  If there
> 	    aren't enough elements to assign one to this target, then no
> 	    assignment is made to this variable, leaving it with whatever its
> 	    previous value was.
> 	? VARIABLE = EXPR
> 	    This is also an optional target, but if there aren't enough list
> 	    elements available to assign one to this target, the result of
> 	    evaluating EXPR is assigned to it instead.  Thus, EXPR is a kind of
> 	    `default value' for the variable.  The default value expressions
> 	    are evaluated and assigned working from left to right *after* all
> 	    of the other assignments have been performed.
> 	@ VARIABLE
> 	    Analogously to MOO argument lists, this variable is assigned a list
> 	    of all of the `leftover' list elements in this part of the list
> 	    after all of the other targets have been filled in.  It will be
> 	    assigned the empty list, if there aren't any elements left over.  I
> 	    call this a `rest' target, since it gets the rest of the elements.
> 	    There may be at most one rest target in TARGETS.
>    If there aren't enough list elements to fill all of the required targets, or
>    if there are more than enough to fill all of the required and optional
>    targets but there isn't a rest target to take the leftover ones, then E_ARGS
>    is raised.
>       Here are some examples of how this works.  Assume first that the verb
>    `me:foo()' contains the following code:
> 		b = c = e = 17;
> 		{a, ?b, ?c = 8, @d, ?e = 9, f} = args;
> 		return {a, b, c, d, e, f};
>    Then the following calls return the given values:
> 	    `me:foo(1) ! ANY'              => E_ARGS
> 	    me:foo(1, 2)                   => {1, 17, 8, {}, 9, 2}
> 	    me:foo(1, 2, 3)		   => {1, 2, 8, {}, 9, 3}
> 	    me:foo(1, 2, 3, 4)		   => {1, 2, 3, {}, 9, 4}
> 	    me:foo(1, 2, 3, 4, 5)	   => {1, 2, 3, {}, 4, 5}
> 	    me:foo(1, 2, 3, 4, 5, 6)	   => {1, 2, 3, {4}, 5, 6}
> 	    me:foo(1, 2, 3, 4, 5, 6, 7)	   => {1, 2, 3, {4, 5}, 6, 7}
> 	    me:foo(1, 2, 3, 4, 5, 6, 7, 8) => {1, 2, 3, {4, 5, 6}, 7, 8}
>    Finally MOO has a convenient mechanism for naming verb arguments, checking
>    for there being exactly the right number of arguments, handling optional and
>    `rest' arguments, etc.  I intend to start every new MOO verb of mine with a
>    scattering assignment of `args', and I encourage other MOO programmers to do
>    the same.
> -------------------------------------------------------------------------------
> 
> 
> 
-----------------------------------------------------------------------------
Kevin L. Kitchens                                   Internet: klk@crl.com
PEI Programming - Stone Mountain, Georgia                CIS: 76614,2161
-----------------------------------------------------------------------------
  "The comments contained within ARE those of my employer...myself!"
-----------------------------------------------------------------------------



Home | Subject Index | Thread Index