|Column Tag:||Basic School
Using the Segment Manager from Basic
By Dave Kelly, MacTutor Editorial Board
This month we'll discuss Macintosh memory management (part 2, part 1 was discussed back in the June 1985 MacTutor). While the information back in June 1985 is still valid, there are some other (new?) considerations that I haven't mentioned yet. If you have the June '85 MacTutor go back and review that information as it will be useful here as we talk about it some more. It would also be helpful to read Inside Macintosh Vol. I chapter 3 on Memory Management and Vol II chapter 2 on the Segment Loader.
Now that we have as much memory available as we ever dreamed we might have (remember when it was so great to have 16K RAM?), we need to talk about the best ways to make use of it. In review, please recall that the key to memory management in MS Basic is the CLEAR statement. Even on a Macintosh Plus you should use CLEAR to make the best use of the available memory. The following procedure will help you to quickly set up the memory configuration:
1. Load your program and before running it type PRINT FRE(0) in the command window to determine the amount of memory used by your program listing. Record this number for later use.
2. Run the program. The intent here is to determine the amount of memory that your variables need. For some programs you may not be able to run enough of the program to declare all of the variables. In that case you may have to do some rough guessing here. Either after running your program or sometime during the program (this may be more convenient) you should print FRE(0) again. The difference between this number and the number you recorded in step 1 above is the amount of memory allocated to the data segment that is used for your variables. (FRE(-1) returns the number of bytes not used in the heap; FRE(-2) returns the number of bytes in the stack that have not been used; you may use these if it will help you).
3. Now it's time to guess. Try to decide if the program will ever need more memory than this in the future. If you need to increase the program size later or allocate more variables you will need to allow some space for this. If you are not sure you may want to just add 20% or so more memory to the length of the program. Whatever you decide here will be your data segment memory.
4. Now, if your program uses any Macintosh resources you will need to figure out how much of the heap to use. If you have extra memory like on the Mac plus it never hurts to have extra heap space. In fact, the heap may be more heavily used than the data segment depending on how many controls, or windows, etc. you may be using. Keep in mind also that if any desk accessories are opened they will need some heap space of their own. By printing FRE(-2) it will show the heap memory allocation. If you print FRE(-2) before running the program you will know the amount of heap available to the program. After the program is run and resources are loaded, the amount of heap will decrease. You need to know the difference between the FRE(-2) before and after running the program. Record this number.
5. Now you need to calculate the stack-size to use in the CLEAR statement. Many times it is not even necessary to change the data segment (leave it as large as the memory you have left). The number to use for the stack-size is calculated by stack-size = Total Ram - (data-segment-size + heap-size).
Fig. 1 Output of our Segment Program
Now how do you know when you need to adjust the memory allocation? Ans: Anytime that you notice an unusual amount of disk activity, especially just before displaying controls (edit fields, buttons, menus, windows etc.). It doesn't matter if you have a Mac Plus or not, if you don't adjust your memory properly, then the memory will not be used efficiently. In order to remain compatible with the 512K & 128K Mac, the heap and stack default sizes are the same on the Mac plus. This may explain why so many Mac programs need to read the disk so often even though you may think you have a lot of memory (1Meg really is plenty for Basic!!). Good memory management planning can improve your program tremendously. Another thing to keep in mind is that once the CLEAR statement has been used to change the memory allocation it will remain that way for the next program that you load into memory. A second CLEAR statement will not reallocate the memory properly. It is best to quit to the finder and start all over unless the new program uses the same memory configuration.
So much for MS Basic memory management. The comments above in addition to those in the June '85 MacTutor should give you all you need to handle most of your memory management. Memory management in other languages is handled through use of the Segment Loader. Access to the Segment Loader is not provided in MS Basic. ZBasic provides the SEGMENT and SEGMENT RETURN statements which use the Segment Loader. We'll talk about those in a moment. First a few words about the Segment Loader. The Segment Loader allows you to divide up the code of your application into several parts. When the Finder starts up your application the Segment Loader is called to load in the main segment of your program. Other segments are loaded in automatically as needed. After a segment is no longer needed, your application can call the segment loader to make the segment purgable. This way if you have only a small amount of memory available you may still run a large program. The maximum size of a segment is 32K. Any code that is not used often may be in a separate segment so that it may be swapped in whenever it is needed. Another function of the Segment Loader is in supplying information from the Finder.
Your ZBasic code will be compiled into code segments (resources of type 'CODE') automatically, no matter if you use the SEGMENT statements or not. When the code reaches a length of about 28K the segment will end and the next segment will be loaded. The SEGMENT RETURN statement marks the end of the CODE segment and returns to the calling routine (cannot be called from the same segment). You should arrange your application such that a segment consists of a subroutine. The routines you decide to separate into segments preferably should be routines which are only used occasionally (example: print routines).
The following example program demonstrates how to "segmentize" your programs. However, because the segments in the example are small (less than 100 bytes), the segment memory swapping is virtually non-existent on a Macintosh plus (and probably all the other size Macs too!). If you really want to try this out, you'll have to write some larger CODE segments and run on a smaller Mac. You can check the segment sizes by choosing the COMPILE option of the ZBasic command menu. The sizes of the segments will be displayed. The remaining memory can be found with the MEM(-1) function. MEM(-1) returns the maximum amount of available memory for the program and variables.
Another part of the Segment Loader is supported by ZBasic with the FINDERINFO function. The FINDERINFO function returns information to your application which tells the application how the application was selected from the Finder. The function returns variables with the number of files selected, the name of the files and volume number, and a message variable indicating if the file should be loaded or printed. This way the files may be printed (or whatever) without going through the full set of initialization routines that your application may have. If no files have been selected then your application should open a new untitled document and continue from there. The syntax is: message%= FINDERINFO (count%, var$[(n)], type%[(n)], volume%[(n)]). The message variable is the result of the FINDERINFO function. If the result is equal to zero then the file should be loaded otherwise the file should be printed. The Count% variable returns the number of files . The variable var$(n) is a string variable or string array with the names of the files which have been selected. The file type of the selected files is stored in a long integer word which must be converted to a string by using the MKI$ statement. The volume% variable is the the disk volume where the file is located.
To get the FINDERINFO function to work, first you must compile your program as an application. There are two or three statements that must be used properly in order to compile the program so that the function will work. First type CREATOR "ffff" where "ffff" is the creator name which will be given to the application. The other statement that must match this one is the DEFOPEN "ffffcccc" statement which sets up the file type ("ffff") and the file creator ("cccc") for the data files created by your application. In the demo program the creator is "DAVE" and the file type is "DATA". The DEFOPEN statement is used in the application program when opening up new data files. Now that the creator is the same for the application as for the data, we need to set the bundle bit with the BUNDLE 1 statement. Type BUNDLE 1 in the command window. Now it is time to compile the program into an application. Select RUN* from the menus to create the application. New data files which are created with the application will now be linked to each other. Selecting one or more data files while in the Finder and selecting PRINT or OPEN will open the application which the data files belong and send info to the application via the FINDERINFO statement. To see the results of this function, you may want to run the demo and create a few sample data files and see what happens when they are selected for printing or to be opened.
If you have problems getting the FINDERINFO information, be sure that you set the creator and bundle bit before compiling the application. You should also be sure to define all the variables in the function before calling it. More information on the Segment Loader is available in Vol 2, Chap. 2 of Inside Macintosh (The Segment Loader chapter).
REM Segment Loader Example ZBasic 3.02
REM ©MacTutor 1986
REM by Dave Kelly
DIM 31 Name$(5),Vol%(5),Type&(5),Mes$(1)
REM up to 5 files can be selected.
Message%=FINDERINFO(Count%, Name$(0), Type&(0), Vol%(0))
WINDOW 1,"Segment Sample",(50,100)-(475,300),257
IF Count%<>0 THEN GOSUB "DisplayFINDERINFO"
MENU 1,1,1,"Create Data File"
MENU 2,1,1,"Main Segment"
MENU 2,2,1,"1st Segment"
MENU 2,3,1,"2nd Segment"
MENU 2,4,1,"3rd Segment"
MENU 2,5,1,"Check memory available and unload unused segments"
ON MENU GOSUB "Menuevent"
ON Menunumber GOSUB "Filemenu","Segmenu"
IF Menuitem=1 THEN "Openfile"
IF Menuitem<>2 THEN RETURN
DEF OPEN "DATADAVE"
Filename$=FILES$(0,"Create new file as ..." ,"Segment DATA")
IF Filename$="" THEN RETURN
PRINT "New file named ";Filename$;" has been created.":PRINT
PRINT "Quit the program and click on "; Filename$; " to see how the"
PRINT "FINDERINFO function works. Select PRINT or OPEN from the Finder."
PRINT "The FINDERINFO function will indicate what has been selected."
PRINT "You may want to create multiple data files and try printing"
PRINT "or opening them to see what happens."
ON Menuitem GOSUB "Main", "Seg1", "Seg2", "Seg3", "Memory"
PRINT "This is part of the main segment"
PRINT "This is part of the 1st segment"
PRINT "This is part of the 2nd segment"
PRINT "This is part of the 3rd segment"
PRINT "Memory = ";X&
Mes$(0)="Data file(s) should be loaded"
Mes$(1)="Data file(s) should be printed"
PRINT "Message% = ";Message%;" Therefore... "; Mes$( Message%)
PRINT "Count% = ";Count%; "file(s) have been passed to this application"
PRINT "Filenames are:"
FOR C=0 TO Count%-1