Difference between revisions of "AGIWiki/Message"
m (→Data types) |
(fixing link) |
||
Line 34: | Line 34: | ||
print("%m2 But he'll be back soon."); | print("%m2 But he'll be back soon."); | ||
</syntax> | </syntax> | ||
The message is stored like this in the logic resource, but when the [[ | The message is stored like this in the logic resource, but when the [[AGIWiki/Interpreter|interpreter]] displays it, it will replace <code>%m2</code> with message 2, so the text "He’s not here. But he’ll be back soon." will be displayed. See [[AGIWiki/Message#Building messages|Building messages]] for more details. | ||
=== Avoiding code duplication === | === Avoiding code duplication === |
Revision as of 07:23, 5 August 2013
A message is a string of text that is used by commands like print to give the user information. Each logic resource can contain up to 255 messages, with message numbers starting from 1 and going to 255. A message can also be as long as you like; however, there are practical limitations to this. See the message too verbose error. Messages can be defined implicitly or explicitly. An explicit message definition uses the #message
command:
<syntax type="C++">
- message 2 "He's not here."
</syntax>
This sets the text of message 2, or m2
, to "He's not here." Messages are implicitly defined when they do not appear in any #message
command but do appear within a print command:
<syntax type="C++">
print("He's not here.");
</syntax>
When a message is implicitly defined in this manner, the compiler will essentially add a #message
command to the logic, but it will be invisible. Although possible, it is not generally practical to find out which message number is assigned to the message. Furthermore, additional implicit assignments in the logic file can change which message number is assigned to it. If you need to know the number of a message for sure, use an explicit message definition.
If you use an explicit message definition such as the one shown previously, you can print the message by either duplicating the message in full or specifying the message number: <syntax type="C++"> print("He's not here."); print(m2); </syntax> As you might expect, there are advantages and disadvantages to both methods of defining messages.
Advantages of implicit message definition
The main advantages of implicit message definition are that it is less code to write and that you don't have to worry about manually assigning a message number to each new message you include in your logic. Because some logics may have many, many print statements (particularly if your game is very responsive to player commands), assigning message numbers can quickly become a nuisance.
Advantages of explicit message definition
Explicit message definition also has some advantages, namely that it allows you to avoid duplicating code and allows you to reference the message from within another message (described shortly). Also, commands like print.v use a message number determined by the value of a variable, so when using these command you sometimes need to know the numbers of certain messages.
Message referencing
You can refer to one message from within another by using message referencing. This is done by including %m and then the message number within the other message. For example: <syntax type="C++">
- message 2 "He's not here."
print("%m2 But he'll be back soon.");
</syntax>
The message is stored like this in the logic resource, but when the interpreter displays it, it will replace %m2
with message 2, so the text "He’s not here. But he’ll be back soon." will be displayed. See Building messages for more details.
Avoiding code duplication
For an example where explicitly defining a message can avoid code duplication, let's take a simple "You're not close enough." message that occurs commonly in AGI games. <syntax type="C++"> if (said("get", "cat")) {
if (!posn(ego, 20, 30, 40, 50)) { print("You're not close enough."); }
}
if (said("get", "cup")) {
if (!posn(ego, 60, 60, 80, 90)) { print("You're not close enough."); }
} </syntax> Obviously here we've got some duplicate code: print("You're not close enough.");. At first glance, using explicit messages can provide a small advantage: <syntax type="C++"> if (said("get", "cat")) {
if (!posn(ego, 20, 30, 40, 50)) { print(m1); }
}
if (said("get", "cup")) {
if (!posn(ego, 60, 60, 80, 90)) { print(m1); }
}
- message 1 "You're not close enough."
</syntax>
Or, using defines: <syntax type="C++">
- define NotCloseEnough m1
if (said("get", "cat")) {
if (!posn(ego, 20, 30, 40, 50)) { print(NotCloseEnough); }
}
if (said("get", "cup")) {
if (!posn(ego, 60, 60, 80, 90)) { print(NotCloseEnough); }
}
- message 1 "You're not close enough."
</syntax>
Obviously, either example saves a few keystrokes, but it also provides another advantage. Suppose you want to change your "not close enough" message to read "You're too far away." In the original code, where the message is implicitly defined, you will have to search and replace in your code for every place where the text "You're not close enough." appears. <syntax type="C++"> if (said("get", "cat")) {
if (!posn(ego, 20, 30, 40, 50)) { print("You're too far away."); // changed }
}
if (said("get", "cup")) {
if (!posn(ego, 60, 60, 80, 90)) { print("You're too far away."); // changed }
} </syntax>
The danger that is not necessarily obvious here is that if you miss one or make a typo, you will accidently create a new implicitly-defined message! <syntax type="C++"> if (said("get", "cat")) {
if (!posn(ego, 20, 30, 40, 50)) { print("You're too far away."); // changed }
}
if (said("get", "cup")) {
if (!posn(ego, 60, 60, 80, 90)) { /* OOPS!! No period at the end, so this is a new message! */
print("You're too far away"); // changed }
} </syntax> Using explicit message definitions allows the change to be made in just one place: the message definition. <syntax type="C++">
- define NotCloseEnough m1
if (said("get", "cat")) {
if (!posn(ego, 20, 30, 40, 50)) { print(NotCloseEnough); }
}
if (said("get", "cup")) {
if (!posn(ego, 60, 60, 80, 90)) { print(NotCloseEnough); }
}
- message 1 "You're too far away." // changed
</syntax>
Building messages
As mentioned in the Message referencing section, you can include messages in other messages. You can also include other things in messages:
%vN
|
value of variable N |
%wN
|
Nth word that the player typed in |
%sN
|
String N |
%mN
|
Message N from current logic (except logic 0) |
%gN
|
Message N from logic 0 |
When using %v
, you can place a |
after the number and then another number which determines the minimum number of digits for the var. For example, if you use %v40|2
and the value of v40
is 8, then the %v40|2
will be replaced with 08. If the value of v40
was 100 or more, all three digits would be displayed.
Example
<syntax type="C++"> v255 = 20; print("The value of v255 is %v255."); </syntax> This code will print the message "The value of v255 is 20."
Advantages
Message inclusion allows you to do things that are not otherwise possible, such as printing a number that is not known when you are creating the game. It also allows you to save considerable amounts of memory.
Take the following code from logic 0 of the AGI Studio Template Game, which informs the player that he or she used a word that the game doesn't understand: <syntax type="C++"> if (input_recieved &&
unknown_word_no > 0) { reset(input_recieved); if (unknown_word_no == 1) { print("I don't understand \"%w1\""); } if (unknown_word_no == 2) { print("\"%w2\" is not in my vocabulary."); } if (unknown_word_no == 3) { print("What is \"%w3\""); } if (unknown_word_no == 4) { print("I don't understand \"%w4\""); } if (unknown_word_no == 5) { print("\"%w5\" is not in my vocabulary."); } if (unknown_word_no == 6) { print("What is \"%w6\""); }
} </syntax>
This code is in fact wasting a few dozen bytes of memory. Although this does not seem like much, in a programming environment like AGI where memory is a precious resource, these bytes can make a difference. When the compiler sees this code, it will generate invisible message definitions similar to the following: <syntax type="C++">
- message 10 "I don't understand \"%w1\""
- message 11 "\"%w2\" is not in my vocabulary."
- message 12 "What is \"%w3\""
- message 13 "I don't understand \"%w4\""
- message 14 "\"%w5\" is not in my vocabulary."
- message 15 "What is \"%w6\""
</syntax> Note: the message numbers were chosen arbitrarily and are not necessarily the message numbers that will be generated by the compiler.
Clearly, there are some parts of these messages that simply cannot be changed (the %w1, %w2, %w3, and so on, which are needed to put the unknown word that the player typed into the message), but there are also some obvious duplications here, namely:
- "I don't understand"
- "is not in my vocabulary"
- "What is"
Each of these phrases appears in two of the messages above. Using explicit message definitions we can reduce the amount of memory used, as shown below: <syntax type="C++"> if (input_recieved &&
unknown_word_no > 0) { reset(input_recieved); if (unknown_word_no == 1) { print("%g10%w1\""); } if (unknown_word_no == 2) { print("\"%w2%g11"); } if (unknown_word_no == 3) { print("%g12%w3\""); } if (unknown_word_no == 4) { print("%g10%w4\""); } if (unknown_word_no == 5) { print("\"%w5%g11"); } if (unknown_word_no == 6) { print("%g12%w6\""); }
}
- message 10 "I don't understand \""
- message 11 "\" is not in my vocabulary."
- message 12 "What is \""
</syntax>
Note: this code uses %g
N because the messages are defined in logic 0.
This code does the exact same thing as the earlier code, but it uses less memory. The obvious disadvantage here is that the code is a little less readable. Unfortunately, this trade-off is somtimes necessary. You can use defines to overcome it if you like. The advantage is that the compiler will now use something similar to the following message definitions: <syntax type="C++">
- message 10 "I don't understand \""
- message 11 "\" is not in my vocabulary."
- message 12 "What is \""
- message 13 "%g10%w1\""
- message 14 "\"%w2%g11"
- message 15 "%g12%w3\""
- message 16 "%g10%w4\""
- message 17 "\"%w5%g11"
- message 18 "%g12%w6\""
</syntax> The next obvious question is how much memory this actually saves. Because the message numbers are chosen at random and do not necessarily reflect what the real message numbers would be, this figure is not necessarily exact (you will have to give or take 6 bytes):
- Avoiding the duplication of the repeated parts of the messages saves 55 bytes.
- The
%g
N portion of each of the six messages uses 24 bytes. - So, the total savings with this code are 31 bytes.
Again, 31 bytes may not seem like much, but 31 bytes saved here and another 31 bytes saved there can add up to avoiding an out of memory error. In this particular code, further memory could be saved by printing the same message any time an unknown word is encountered instead of using three different variations. Obviously, this reduces the variety of responses to unknown words, but if you are running low on memory then this might be an acceptable trade-off.
Note: again, 31 bytes is not an exact figure. For technical reasons, each message adds about 3 additional bytes to the logic resource beyond the number of bytes required for the message itself (which is the same as the number of different characters in the message). So, the actual savings here are probably more along the lines of 22 bytes, give or take 6.
Sources
- AGI Studio help file