1. An elegant solution?
- Posted by Michael Sabal <mjs at OSA.ATT.NE.JP>
Feb 02, 1999
-
Last edited Feb 03, 1999
OK, some people want structures or records or whatever, but implementing =
them would break code. Let's try this on for size, taking inspiration =
from routine_id. Every sequence, whether top level or sublevel, has at =
least one attribute flag for the length of the sequence. Let's add =
another flag called "locked". This flag is boolean, so it won't really =
use any significant memory. Whenever a sequence is created, locked =3D =
false. So far, everything is backwards compatible. Now, we add two new =
procedures and two functions to the internal library. For example:
sequence point3D
This sets the locked attribute to true and initializes the three =
positions to garbage (or 0). Now, in append, prepend, &, and left-value =
of point3D, the change fails, generating an error. If you set an atom =
as "Hello", you get an error, so this would appear as the same. So,
point3D =3D {24,44,49} -- this is OK
point3D =3D {24,44,49,40} -- checks if sequence - OK; check if locked - =
yes; check locked length - 3;
-- check if assignment is the right length - NO, so error. If yes, then =
check if each element is the right
-- type. If no, then type-check error.
point3D =3D {24,44} -- no good, same as above.
point3D[2] =3D 44 -- this is OK
point3D[3] =3D 44.5 -- generates a type-check failure after checking for =
locked status.
point3D =3D append(point3D,...)=20
point3D =3D prepend(point3D,...)
point3D =3D point3D & ...
-- All these generate errors for trying to change the length of a locked =
sequence.
To unlock the sequence and return it to normal operations,
unlock_sequence(point3D)
To check its locked status,
if is_locked(point3D) then
...
end if
And to check the validity of elements of a sequence of locked sequences, =
we would have to use
if equals_sequence(point3Dgrid[1],{"integer","integer","integer"}) then
... -- returns 1 if true, 0 if false
end if
Would this be an acceptable / elegant solution? It would not define any =
new types, shouldn't break existing code, and would allow a Euphorized =
structure syntax. Ideas?
Michael J. Sabal
mjs at osa.att.ne.jp
http://home.att.ne.jp/gold/mjs/index.html
2. Re: An elegant solution?
Although I'm not very intelligent, I understood everything.
So I think it is the elegant solution.
You could also write:
global function structure(sequence s)
return is_locked(s)
end function
.... if structure(foo) then
puts(1,"you're right")
endif
but there's no "." or "::" .
(on my french keyboard it's faster to type 2 ":" than "shift-." 
----------
From: Michael Sabal
To: EUPHORIA at LISTSERV.MUOHIO.EDU
Subject: An elegant solution?
Date: Tuesday 2 February 1999 23:49
<deleted...>
To check its locked status,
if is_locked(point3D) then
...
end if
3. An elegant solution?
My e-mailer has had some trouble lately, and nobody who's still talking =
about structures has addressed this suggestion, so I'm sending it again.
>OK, some people want structures or records or whatever, but =
implementing them would break code. Let's try this on for size, taking =
inspiration from routine_id. Every sequence, whether top level or =
sublevel, has at least one attribute flag for the length of the =
sequence. Let's add another flag called "locked". This flag is =
boolean, so it won't really use any significant memory. Whenever a =
sequence is created, locked =3D false. So far, everything is backwards =
compatible. Now, we add two new procedures and two functions to the =
internal library. For example:
sequence point3D
This sets the locked attribute to true and initializes the three =
positions to garbage (or 0). Now, in append, prepend, &, and left-value =
of point3D, the change fails, generating an error. If you set an atom =
as "Hello", you get an error, so this would appear as the same. So,
point3D =3D {24,44,49} -- this is OK
point3D =3D {24,44,49,40} -- checks if sequence - OK; check if locked - =
yes; check locked length - 3;
-- check if assignment is the right length - NO, so error. If yes, then =
check if each element is the right
-- type. If no, then type-check error.
point3D =3D {24,44} -- no good, same as above.
point3D[2] =3D 44 -- this is OK
point3D[3] =3D 44.5 -- generates a type-check failure after checking for =
locked status.
point3D =3D append(point3D,...)=20
point3D =3D prepend(point3D,...)
point3D =3D point3D & ...
-- All these generate errors for trying to change the length of a locked =
sequence.
To unlock the sequence and return it to normal operations,
unlock_sequence(point3D)
To check its locked status,
if is_locked(point3D) then
...
end if
And to check the validity of elements of a sequence of locked sequences, =
we would have to use
if equals_sequence(point3Dgrid[1],{"integer","integer","integer"}) then
... -- returns 1 if true, 0 if false
end if
Would this be an acceptable / elegant solution? It would not define any =
new types, shouldn't break existing code, and would allow a Euphorized =
structure syntax. Ideas?
Michael J. Sabal
mjs at osa.att.ne.jp
http://home.att.ne.jp/gold/mjs/index.html
