* Difficulty compiling ada program!
@ 1998-07-14 0:00 Ramsey Khair
1998-07-14 0:00 ` Markus Kuhn
` (2 more replies)
0 siblings, 3 replies; 4+ messages in thread
From: Ramsey Khair @ 1998-07-14 0:00 UTC (permalink / raw)
Hi,
Can sombody please try compiling this source code as it keep saying "End of
file expected" you have spilt the files where it says, if anybody has any
luck please let me know!
Ramsey
--::::::::::
--abstract.412
--::::::::::
--ASSET_A_412: Event Set Manager Package
-- This package allows for managing pending events for a simulation
model.
--There are functions to get a new event (either from the free list or to
--allocate a new event), to free an event, to schedule a future event to
enter
--the event set, and to get the next event from the event set. Event
records are
--used to hold all information about a pending event in a simulation model,
--including the time of the event, where it is to happen, and what will
happen.
--Event records are always dynamically allocated and manipulated with
pointers.
--In many simulation models, their number will vary dynamically during the
life
--of the model. All aspects of event-set structure are hidden;
conceptually,
--the event set may be viewed as a linked list ordered by the time at which
--events are scheduled to happen. For the sake of efficiency, a more
complex
--data structure, called a splay tree, is used.
-- The generic pending event set manager is a transliteration of a
Pascal
--version Doug Jones wrote some years ago, properly packaged in Ada style to
--enforce information hiding. The same Pascal source has been
transliterated
--into C by David Brower, and his C code was used as the basis for a C++
--version that is now in the GNU C++ standard distribution. The Pascal
--code has been bug free for a number of years now, and has been widely
--used; the C code had only one bug in its first release, and it has been
--bug free for years.
--::::::::::
--eventset.ada
--::::::::::
-- package GENERIC_EVENT_HANDLER is a generic event set management package.
-- Written by Douglas Jones.
-- Delete and Splay operations added by Bindhu Radhadevi.
-- The implementation used here is based on the splay-tree abstract data
type,
-- as implemented in Pascal and tested in:
-- An Empirical Comparison of Priority-Queue and Event-Set
Implementations,
-- by Douglas W. Jones, Comm. ACM 29, 4 (Apr. 1986) 300-311.
-- The basic splay tree algorithms were originally presented in:
-- Self Adjusting Binary Trees,
-- by D. D. Sleator and R. E. Tarjan, Proc. ACM SIGACT Symposium on
Theory
-- of Computing (Boston, Apr 1983) 235-245.
generic
type AUX_TYPE is private;
-- The type of the application dependent field of an event record.
-- Usually, this is a record type.
type TIME_TYPE is private;
with function ">"(U, V: TIME_TYPE) return BOOLEAN is <>;
-- The type of the time field of an event-record, and its ordering
-- operator. Ordinarily, this will be a numeric type, and the
-- comparison operator will be found implicitly and not explicitly
-- passed as a generic parameter.
package GENERIC_EVENT_HANDLER is
type EVENT;
-- Event records are used to hold all information about a pending
-- event in a simulation model, including the time of the event
-- where it is to happen, and what will happen.
type EVENT_REF is access EVENT;
-- Event records are always dynamically allocated and manipulated
-- with pointers. In many simulation models, their number will
-- vary dynamically during the life of the model.
type LINK_FIELDS is limited private;
-- All aspects of event-set structure are hidden here; conceptually,
-- the event set may be viewed as a linked list ordered by the time
-- at which events are scheduled to happen. For the sake of
-- efficiency, a more complex data structure is used.
type EVENT is record
-- Private fields:
LINK: LINK_FIELDS;
-- Public fields needed by EVENT_HANDLER:
TIME: TIME_TYPE;
-- other fields may be added here (as a generic parameter):
AUX: AUX_TYPE;
end record;
function EMPTY return BOOLEAN;
-- Test to see if the event-set is empty.
function NEW_EVENT return EVENT_REF;
-- Get an event-record from the free-list or allocate a new one.
-- May raise STORAGE_ERROR if memory is exhausted.
procedure FREE_EVENT( N: in out EVENT_REF );
-- Deallocate an event-record, adding it to the free-list.
-- May raise CONSTRAINT_ERROR if N is null.
-- May raise CONSTRAINT_ERROR if N is in the free-list or event-set.
procedure SCHEDULE( N: in EVENT_REF );
-- Add the event N to the event-set.
-- May raise CONSTRAINT_ERROR if N is null.
-- May raise CONSTRAINT_ERROR if N is in the free-list or event-set.
procedure GET_NEXT( N: out EVENT_REF );
-- Remove and return an event from the event-set. The event
-- returned will be one of the events that has a TIME field less
-- than or equal to the time fields of all other events in the
-- event-set.
procedure DELETE( N: in EVENT_REF);
-- Delete an event N from the event set.
-- May raise CONSTRAINT_ERROR if N is not in event set.
EVENT_SET_EMPTY: exception;
-- raised by GET_NEXT on an empty event set.
private
-- The data structures defined here are specific to the splay-tree
-- implementation and may need to be changed for other implementations.
type LINK_FIELDS is record
LEFT: EVENT_REF;
-- Left child of the event record in the event-set.
-- Successor pointer for events in the free-list.
-- For debugging purposes only, X.LEFT = X indicates an event
-- that is outside the free-list or event-set.
RIGHT: EVENT_REF;
-- Right child of event record in the event-set.
UP: EVENT_REF;
-- Parent of event record in the event-set.
end record;
end GENERIC_EVENT_HANDLER;
package body GENERIC_EVENT_HANDLER is
EVENT_SET: EVENT_REF := null;
-- This is the root of the binary tree representing the event-set.
FREE_LIST: EVENT_REF := null;
-- This is the head of the linked list holding free EVENT records.
DEBUG: constant BOOLEAN := TRUE;
-- Controls runtime checks on preconditions to operations. Set to
-- false for the same reason you would use the pragma SUPPRESS.
-- This is included on the assumption that the compiler will remove
-- dead code if DEBUG is false.
function EMPTY return BOOLEAN is
begin
return EVENT_SET = null;
end EMPTY;
function NEW_EVENT return EVENT_REF is
N: EVENT_REF;
begin
if FREE_LIST = null then
N := new EVENT;
else
N := FREE_LIST;
FREE_LIST := FREE_LIST.LINK.LEFT;
end if;
if DEBUG then
N.LINK.LEFT := N; -- flag event as being outside the free list
end if;
return N;
end NEW_EVENT;
procedure FREE_EVENT( N: in out EVENT_REF ) is
begin
if DEBUG then -- assertions
if N = null then raise CONSTRAINT_ERROR; end if;
-- N must be an event record
if N.LINK.LEFT /= N then raise CONSTRAINT_ERROR; end if;
-- N must be outside the event-set
end if;
N.LINK.LEFT := FREE_LIST;
FREE_LIST := N;
end FREE_EVENT;
procedure SCHEDULE( N: in EVENT_REF ) is
LEFT: EVENT_REF; -- the rightmost event in the left tree
RIGHT: EVENT_REF; -- the leftmost event in the right tree
NEXT: EVENT_REF; -- the root of that part of the tree not yet
split
TEMP: EVENT_REF;
TIME: TIME_TYPE; -- the time of event N
begin
if DEBUG then -- assertions
if N = null then raise CONSTRAINT_ERROR; end if;
-- N must be an event record
if N.LINK.LEFT /= N then raise CONSTRAINT_ERROR; end if;
-- N must be outside the event-set
end if;
N.LINK.UP := null;
NEXT := EVENT_SET;
EVENT_SET := N;
if NEXT = null then -- trivial schedule operation
N.LINK.LEFT := null;
N.LINK.RIGHT := null;
return;
else -- difficult schedule operation
TIME := N.TIME;
LEFT := N;
RIGHT := N;
-- N's left and right children will hold the right and left
-- splayed trees resulting from splitting NEXT around TIME,
-- but note that the left and right children will be reversed
if NEXT.TIME > TIME then
goto WALK_LEFT;
end if;
<>
-- assert NEXT.TIME <= TIME
loop -- walk to the right in the left tree
TEMP := NEXT.LINK.RIGHT;
if TEMP = null then
LEFT.LINK.RIGHT := NEXT;
NEXT.LINK.UP := LEFT;
RIGHT.LINK.LEFT := null;
goto END_WALK; -- job done, entire tree is split
end if;
if TEMP.TIME > TIME then
LEFT.LINK.RIGHT := NEXT;
NEXT.LINK.UP := LEFT;
LEFT := NEXT;
NEXT := TEMP;
goto WALK_LEFT; -- change sides
end if;
NEXT.LINK.RIGHT := TEMP.LINK.LEFT;
if TEMP.LINK.LEFT /= null then
TEMP.LINK.LEFT.LINK.UP := NEXT;
end if;
LEFT.LINK.RIGHT := TEMP;
TEMP.LINK.UP := LEFT;
TEMP.LINK.LEFT := NEXT;
NEXT.LINK.UP := TEMP;
LEFT := TEMP;
NEXT := TEMP.LINK.RIGHT;
if NEXT = NULL then
RIGHT.LINK.LEFT := null;
goto END_WALK; -- job done, entire tree is split
end if;
exit when NEXT.TIME > TIME; -- change sides
end loop;
<>
-- assert NEXT.TIME > TIME
loop -- walk to the left in the right tree
TEMP := NEXT.LINK.LEFT;
if TEMP = null then
RIGHT.LINK.LEFT := NEXT;
NEXT.LINK.UP := RIGHT;
LEFT.LINK.RIGHT := null;
goto END_WALK; -- job done, entire tree is split
end if;
if not (TEMP.TIME > TIME) then
RIGHT.LINK.LEFT := NEXT;
NEXT.LINK.UP := RIGHT;
RIGHT := NEXT;
NEXT := TEMP;
goto WALK_RIGHT; -- change sides
end if;
NEXT.LINK.LEFT := TEMP.LINK.RIGHT;
if TEMP.LINK.RIGHT /= null then
TEMP.LINK.RIGHT.LINK.UP := NEXT;
end if;
RIGHT.LINK.LEFT := TEMP;
TEMP.LINK.UP := RIGHT;
TEMP.LINK.RIGHT := NEXT;
NEXT.LINK.UP := TEMP;
RIGHT := TEMP;
NEXT := TEMP.LINK.LEFT;
if NEXT = null then
LEFT.LINK.RIGHT := null;
goto END_WALK; -- job done, entire tree is split
end if;
exit when not (NEXT.TIME > TIME); -- change sides
end loop;
goto WALK_RIGHT;
<>
TEMP := N.LINK.LEFT;
N.LINK.LEFT := N.LINK.RIGHT;
N.LINK.RIGHT := TEMP;
return;
end if;
-- control should never reach this point!
raise CONSTRAINT_ERROR;
end SCHEDULE;
procedure GET_NEXT( N: out EVENT_REF ) is
NEXT: EVENT_REF; -- then next event to deal with
LEFT: EVENT_REF; -- the left child of NEXT
FAR_LEFT: EVENT_REF; -- the left child of LEFT
FAR_FAR_LEFT: EVENT_REF; -- the left child of FAR_LEFT
begin
if EVENT_SET = null then
raise EVENT_SET_EMPTY;
else
NEXT := EVENT_SET;
LEFT := NEXT.LINK.LEFT;
if LEFT = null then -- trivial case, return the root
EVENT_SET := NEXT.LINK.RIGHT;
if EVENT_SET /= null then
EVENT_SET.LINK.UP := null;
end if;
if DEBUG then
NEXT.LINK.LEFT := NEXT; -- mark event as idle
end if;
N := NEXT;
return;
else -- nontrivial case, must find leftmost child of root
loop
-- NEXT is not it; LEFT is not NIL and might be it
FAR_LEFT := LEFT.LINK.LEFT;
if FAR_LEFT = null then
NEXT.LINK.LEFT := LEFT.LINK.RIGHT;
if LEFT.LINK.RIGHT /= null then
LEFT.LINK.RIGHT.LINK.UP := NEXT;
end if;
if DEBUG then
LEFT.LINK.LEFT := LEFT; -- mark event as idle
end if;
N := LEFT;
return;
end if;
-- NEXT and LEFT are not it, FAR_LEFT non-null, could be
it
FAR_FAR_LEFT := FAR_LEFT.LINK.LEFT;
if FAR_FAR_LEFT = null then
LEFT.LINK.LEFT := FAR_LEFT.LINK.RIGHT;
if FAR_LEFT.LINK.RIGHT /= null then
FAR_LEFT.LINK.RIGHT.LINK.UP := LEFT;
end if;
if DEBUG then
FAR_LEFT.LINK.LEFT := FAR_LEFT; -- mark as idle
end if;
N := FAR_LEFT;
return;
end if;
-- NEXT, LEFT, FAR_LEFT are not null and not it, rotate!
NEXT.LINK.LEFT := FAR_LEFT;
FAR_LEFT.LINK.UP := NEXT;
LEFT.LINK.LEFT := FAR_LEFT.LINK.RIGHT;
if FAR_LEFT.LINK.RIGHT /= null then
FAR_LEFT.LINK.RIGHT.LINK.UP := LEFT;
end if;
FAR_LEFT.LINK.RIGHT := LEFT;
LEFT.LINK.UP := FAR_LEFT;
NEXT := FAR_LEFT;
LEFT := FAR_FAR_LEFT;
end loop;
end if;
end if;
-- control should never reach this point!
raise CONSTRAINT_ERROR;
end GET_NEXT;
---- SPLAY is not a visible component of the package, but is
---- only used by DELETE.
procedure SPLAY( N: in EVENT_REF) is
UP: EVENT_REF; -- points to the event currently being dealt with
PREV: EVENT_REF; -- a child of up which has already been dealt
with
UPUP: EVENT_REF; -- the parent of up
UPUPUP: EVENT_REF; -- the grandparent of up
LEFT: EVENT_REF; -- the top of the left subtree being built
RIGHT: EVENT_REF; -- the top of the right subtree beign built
begin
LEFT := N.LINK.LEFT;
RIGHT := N.LINK.RIGHT;
PREV := N;
UP := PREV.LINK.UP;
WHILE UP /= null loop
-- walk up the tree towards the root, splaying all to the left
of
-- N into the left subtree, all to right into the right subtree
UPUP := UP.LINK.UP;
if UP.LINK.LEFT = PREV then -- up is to the right of n
if UPUP /= null then
if UPUP.LINK.LEFT = UP then -- rotate
UPUPUP := UPUP.LINK.UP;
UPUP.LINK.LEFT := UP.LINK.RIGHT;
if UPUP.LINK.LEFT /= null then
UPUP.LINK.LEFT.LINK.UP := UPUP;
end if;
UP.LINK.RIGHT := UPUP;
UPUP.LINK.UP := UP;
if UPUPUP = null then
EVENT_SET := UP;
elsif UPUPUP.LINK.LEFT = UPUP then
UPUPUP.LINK.LEFT := UP;
else
UPUPUP.LINK.RIGHT:= UP;
end if;
UP.LINK.UP := UPUPUP;
UPUP := UPUPUP;
end if;
end if;
UP.LINK.LEFT := RIGHT;
if RIGHT /= null then
RIGHT.LINK.UP := UP;
end if;
RIGHT := UP;
else -- up is to the left of n
if UPUP /=null then
if UPUP.LINK.RIGHT = UP then -- rotate
UPUPUP := UPUP.LINK.UP;
UPUP.LINK.RIGHT := UP.LINK.LEFT;
if UPUP.LINK.RIGHT /= null then
UPUP.LINK.RIGHT.LINK.UP:= UPUP;
end if;
UP.LINK.LEFT := UPUP;
UPUP.LINK.UP := UP;
if UPUPUP = null then
EVENT_SET := UP;
elsif UPUPUP.LINK.RIGHT = UPUP then
UPUPUP.LINK.RIGHT := UP;
else
UPUPUP.LINK.LEFT := UP;
end if;
UP.LINK.UP := UPUPUP;
UPUP := UPUPUP;
end if;
end if;
UP.LINK.RIGHT := LEFT;
if LEFT /= null then
LEFT.LINK.UP := UP;
end if;
LEFT := UP;
end if;
PREV := UP;
UP := UPUP;
end loop;
if EVENT_SET /= PREV then
-- either N wasn't in the event set or the event set is corrupt.
raise CONSTRAINT_ERROR;
end if;
N.LINK.LEFT := LEFT;
N.LINK.RIGHT := RIGHT;
if LEFT /= null then
LEFT.LINK.UP := N;
end if;
if RIGHT /= null then
RIGHT.LINK.UP := N;
end if;
EVENT_SET := N;
N.LINK.UP := null;
end SPLAY;
procedure DELETE( N: in EVENT_REF) is
-- N is deleted from EVENT_SET; the resulting splay tree
-- has been splayed around its new root, which is the successor of N
X: EVENT_REF;
begin
if DEBUG then -- assertions
if N = null then raise CONSTRAINT_ERROR; end if;
-- N must be an event record
if N.LINK.LEFT = N then raise CONSTRAINT_ERROR; end if;
-- N must be in the event-set
-- Note! This is a weak assertion -- the exception won't
-- be raised if N is in the free list; the raise
-- statement in SPLAY will detect that case.
end if;
SPLAY(N);
EVENT_SET := N.LINK.RIGHT;
if EVENT_SET /= null then
EVENT_SET.LINK.UP := null;
GET_NEXT(X);
X.LINK.UP := null;
X.LINK.LEFT := N.LINK.LEFT;
X.LINK.RIGHT := EVENT_SET;
if X.LINK.LEFT /= null then
X.LINK.LEFT.LINK.UP := X;
end if;
if X.LINK.RIGHT /= null then
X.LINK.RIGHT.LINK.UP := X;
end if;
EVENT_SET := X;
else
EVENT_SET := N.LINK.LEFT;
if EVENT_SET /= null then
EVENT_SET.LINK.UP := null;
end if;
end if;
if DEBUG then
N.LINK.LEFT := N; -- flag event as being outside the free list
end if;
end DELETE;
end GENERIC_EVENT_HANDLER;
--::::::::::
--testevnt.ada
--::::::::::
with TEXT_IO;
with GENERIC_EVENT_HANDLER;
procedure EVENT_HANDLER_TEST is
package EVENT_HANDLER is new GENERIC_EVENT_HANDLER( AUX_TYPE => BOOLEAN,
TIME_TYPE => FLOAT );
use EVENT_HANDLER;
use TEXT_IO;
package INT_I_O is new INTEGER_IO(INTEGER);
use INT_I_O;
begin
TEXT_IO.PUT_LINE( "Event handler test!" );
if not EMPTY then
TEXT_IO.PUT_LINE( " -- initialization failure" );
else
TEXT_IO.PUT_LINE( " -- initialization success" );
declare -- test 1, see if one event can be put in and gotten back
A, B: EVENT_REF;
begin
A := NEW_EVENT;
A.TIME := 0.0;
SCHEDULE( A );
if EMPTY then
TEXT_IO.PUT_LINE( " -- SCHEDULE failure" );
else
GET_NEXT( B );
if A = B then
TEXT_IO.PUT_LINE( " -- GET_NEXT got scheduled event" );
else
TEXT_IO.PUT_LINE( " -- GET_NEXT failure" );
end if;
FREE_EVENT( B );
end if;
end;
declare -- test 2, see if it works as a FIFO with times in order
A: EVENT_REF;
begin
for I in 1..1000 loop
A := NEW_EVENT;
A.TIME := FLOAT(I);
SCHEDULE( A );
end loop;
for I in 1..1000 loop
GET_NEXT( A );
if A.TIME /= FLOAT(I) then
TEXT_IO.PUT_LINE( " -- GET_NEXT out of order (1)! " );
end if;
FREE_EVENT( A );
end loop;
end;
declare -- test 3, see if it works as a LIFO with times reversed
A: EVENT_REF;
begin
for I in reverse 1..1000 loop
A := NEW_EVENT;
A.TIME := FLOAT(I);
SCHEDULE( A );
end loop;
for I in 1..1000 loop
GET_NEXT( A );
if A.TIME /= FLOAT(I) then
TEXT_IO.PUT_LINE( " -- GET_NEXT out of order (2)! " );
end if;
FREE_EVENT( A );
end loop;
end;
declare -- test 4, see if it works with times a bit random
A: EVENT_REF;
T: FLOAT;
TABLE : array(1..1000) of EVENT_REF;
begin
for I in 1..1000 loop -- schedule 1000 random events
A := NEW_EVENT;
A.TIME := FLOAT((I * 421 + 2073) mod 9317);
SCHEDULE( A );
end loop;
for I in 1..1000 loop -- schedule 1000 random events
A := NEW_EVENT;
A.TIME := FLOAT(((I+1000) * 421 + 2073) mod 9317);
TABLE(I) := A;
SCHEDULE( A );
end loop;
for I in 1..1000 loop -- delete the second 1000 events
DELETE(TABLE(I));
end loop;
T := -1.0;
GET_NEXT( A );
for I in 2..1000 loop -- dequeue the first 1000 events
if A.TIME < T then
TEXT_IO.PUT_LINE( " -- GET_NEXT out of order (3)! " );
end if;
T := A.TIME;
GET_NEXT( A );
FREE_EVENT( A );
end loop;
end;
if not EMPTY then
TEXT_IO.PUT_LINE( " -- failure, event-set not empty at end" );
else
TEXT_IO.PUT_LINE( " -- success, event-set empty at end" );
end if;
end if;
end EVENT_HANDLER_TEST;
^ permalink raw reply [flat|nested] 4+ messages in thread
* Re: Difficulty compiling ada program!
1998-07-14 0:00 Difficulty compiling ada program! Ramsey Khair
@ 1998-07-14 0:00 ` Markus Kuhn
1998-07-14 0:00 ` Dale Stanbrough
1998-07-14 0:00 ` David C. Hoos, Sr.
2 siblings, 0 replies; 4+ messages in thread
From: Markus Kuhn @ 1998-07-14 0:00 UTC (permalink / raw)
Ramsey Khair wrote:
> Can sombody please try compiling this source code as it keep saying "End of
> file expected" you have spilt the files where it says, if anybody has any
> luck please let me know!
Assuming you are using GNAT, you clearly have not read the manual:
http://www.cl.cam.ac.uk/~mgk25/volatile/gnat_ug.html#SEC9
http://www.cl.cam.ac.uk/~mgk25/volatile/gnat_ug.html#SEC67
You have to split your file up into an *.ads and an *.adb file,
and gnatchop can do this for you very easily as explained in the
GNAT User Guide.
Markus
--
Markus G. Kuhn, Security Group, Computer Lab, Cambridge University, UK
email: mkuhn at acm.org, home page: <http://www.cl.cam.ac.uk/~mgk25/>
^ permalink raw reply [flat|nested] 4+ messages in thread
* Re: Difficulty compiling ada program!
1998-07-14 0:00 Difficulty compiling ada program! Ramsey Khair
1998-07-14 0:00 ` Markus Kuhn
@ 1998-07-14 0:00 ` Dale Stanbrough
1998-07-14 0:00 ` David C. Hoos, Sr.
2 siblings, 0 replies; 4+ messages in thread
From: Dale Stanbrough @ 1998-07-14 0:00 UTC (permalink / raw)
Step 2 is to always, always, always remove bits of the code until you
have the _smallest_ sample of code that replicates the problem.
I certainly know that -my- eyes glazed over when looking at your posting.
Dale
^ permalink raw reply [flat|nested] 4+ messages in thread
* Re: Difficulty compiling ada program!
1998-07-14 0:00 Difficulty compiling ada program! Ramsey Khair
1998-07-14 0:00 ` Markus Kuhn
1998-07-14 0:00 ` Dale Stanbrough
@ 1998-07-14 0:00 ` David C. Hoos, Sr.
2 siblings, 0 replies; 4+ messages in thread
From: David C. Hoos, Sr. @ 1998-07-14 0:00 UTC (permalink / raw)
Ramsey Khair wrote in message
<900407099.23331.0.nnrp-10.c2de523e@news.demon.co.uk>...
>Hi,
>
>Can sombody please try compiling this source code as it keep saying "End of
>file expected" you have spilt the files where it says, if anybody has any
>luck please let me know!
>
Luck has nothing at all to do with it. It just takes common sense and
perseverance.
First, when submitting questions to the group, it is always best to specifiy
the compiler and platform being used, with version numbers for both.
Second, when submitting source code, you should insure that your
mailer/poster software does not split lines, either by setting its
parameters appropriately, or by editing the source so its lines are within
your mailer/poster's capabilites. A number of your comment lines were split
by your mailer/poster software.
Since you did not specify the compiler you are using, I used gnat-3.10p for
Windows NT -- my platform version is 4.0 (SP3).
Here, I followed the instructions in the GNAT User's Guide, particularly the
section entitled "Handling Files with Multiple Units".
This resulted in the file being split into three source files named
appropriately for GNAT. It was in this step that the problem of
inappropriately split comment lines was revealed.
After rejoining the spli comment lines,then there were three compilation
errors in the file "generic_event_handler.adb", where there were lines
containing only "<>".
Initially I commented these lines out, and attempted a recompile (using the
command "gnatmake event_handler_test" to compile, bind and link all required
files with a single command, given that a cursory examination of the source
code suggested that "event_handler_test" was the main program).
Now, the compiler complained about three targets of "goto" statements being
undefined. Knowing that goto labels are of the form ""<<Label_Identifier>>,
and given the three commented-out "<>" lines, I surmised thaat somehow these
labels had been corrupted. Fortunately, the code immediately following two
of these three lines was commented appropriately so that the correct
placement of the three missing labels "Walk_Left", "Walk_Right", and
"End_Walk" could be deduced.
After typing in the three labels, the program compiled error- and
warning-free, and executed correctly.
Normally, I would not take the time nor make the effort I did with this one,
except that I am always on the lookout for possibly re-usable code to
include in my sanitized code repository. This code is Ada83, and as such,
could benefit from some of the new cpapbilities of Ada95. For example, I
would replace heap allocation with a user-defined storage pool to avoid the
potential of heap fragmentation, and the generally poorer performance of
heap allocators over user-defined storage pools. General-purpose components
for simulations should not introduce the possibility that the simulation
could not run for extended periods of time (months or years) without
failure.
In short, "reusable" code from repositories often isn't, and there is no
"silver bullet". When all else fails (but before posting to the group) READ
(AND FOLLOW) THE INSTRUCTIONS.
David C. Hoos, Sr.
^ permalink raw reply [flat|nested] 4+ messages in thread
end of thread, other threads:[~1998-07-14 0:00 UTC | newest]
Thread overview: 4+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
1998-07-14 0:00 Difficulty compiling ada program! Ramsey Khair
1998-07-14 0:00 ` Markus Kuhn
1998-07-14 0:00 ` Dale Stanbrough
1998-07-14 0:00 ` David C. Hoos, Sr.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox