dos_compilers/Borland Turbo C v1

                        WELCOME TO TURBO C
                       ÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄ

    This README file contains information that will be useful and/or of
    interest to you. Please read it in its entirety, referring to it
    when you encounter problems not addressed in the Owner's Handbook.

    TABLE OF CONTENTS
    ÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄ

    1. How to get Help
    2. Corrections/Additions to the Manual
    3. Important tips
    4. Files on the disk
    5. Using Turbo C with DOS 3.2 and a Floating Point Coprocessor
    6. How to compile MicroCalc


    1. HOW TO GET HELP
    ÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄ

    If you have any problems, please read this file and the Owner's
    Handbook first. If you still have a question and need assistance,
    help is available from the following sources:

       1. Type GO BORLAND on the CompuServe bulletin board system
          for instant access to the Borland forums with their
          libraries of technical information and answers to
          common questions.

          If you are not a member of CompuServe, see the enclosed
          special offer, and write for full details on how to
          receive a free IntroPak containing a $15 credit toward
          your first month's on-line charges.

       2. Check with your local software dealer or users' group.

       3. Write to us at the following address:

            Borland International
            Turbo C Technical Support
            4585 Scotts Valley Dr
            Scotts Valley, CA  95066

          Please remember to include your serial number or we
          will be unable to process your letter.

       4. If you have an urgent problem that cannot wait and you have
          sent in the license agreement from the front of your manual,
          you may call the Borland Technical Support department. Please
          have the following information ready before calling:

          a. Product name and serial number on your original
             distribution disk.  Please have your serial number
             ready or we will be unable to process your call.

          b. Product version number. The version number for Turbo C is
             displayed when you first load the program and before you
             press any keys.

          c. Computer brand, model, and the brands and model numbers of
             any additional hardware.

          d. Operating system and version number. (The version number
             can be determined by typing VER at the DOS prompt.)

          e. Contents of your AUTOEXEC.BAT file.

          f. Contents of your CONFIG.SYS file.


    2. CORRECTIONS/ADDITIONS TO THE MANUAL
    ÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄ

    User's Guide (Volume I)

    - Page 10 - The Distribution List is inaccurate.  The correct
      list of files included with Turbo C is given in this readme
      in Section Four.

    - Page 98 - The example has a statement: 
           while ((ch=getch()) != '\n') {
      The correct statement is:
           while ((ch=getch()) != '\r') {
      Note the '\n' is changed to '\r'.

    Reference Guide (Volume II)

    - Page 29 - The description for assert says that to disable
      the assert macro, the symbol NDEBUG must be defined before the
      assert macro is used. In fact it must be placed before
      assert.h is included:

        #define NDEBUG

        #include <assert.h>

    - Page 46 - The correct usage for the bsearch function is:
        void *bsearch(void *key, void *base, int nelem,
                      int width, int (*fcmp)());
      That is, nelem is an int, not a pointer to an int, as
      given in the documentation.

    - Page 108 - The Portability paragraph for the getc function
      says that the functions described in that section (getc,
      getchar, fgetc, etc.) are all in Unix. In fact, all but getch,
      getche, and ungetch are in Unix; these others are MSDOS
      specific.

    - Page 181 - The printf chart is incorrect. The correct chart is:

      prefix  6d      6o      8x      10.2e           10.2f
      %-+#0 |+555   |01053  |0x22b    |+5.50e+000 |+5.50      |
       %-+# |+555   |01053  |0x22b    |+5.50e+000 |+5.50      |  
       %-+0 |+555   |1053   |22b      |+5.50e+000 |+5.50      |  
        %-+ |+555   |1053   |22b      |+5.50e+000 |+5.50      |
       %-#0 |555    |01053  |0x22b    |5.50e+000  |5.50       |  
        %-# |555    |01053  |0x22b    |5.50e+000  |5.50       |  
        %-0 |555    |1053   |22b      |5.50e+000  |5.50       |
         %- |555    |1053   |22b      |5.50e+000  |5.50       |
       %+#0 |+00555 |001053 |0x00022b |+5.50e+000 |+000005.50 |
        %+# |  +555 | 01053 |   0x22b |+5.50e+000 |     +5.50 |
        %+0 |+00555 |001053 |0000022b |+5.50e+000 |+000005.50 |
         %+ |  +555 |  1053 |     22b |+5.50e+000 |     +5.50 |
        %#0 |000555 |001053 |0x00022b |05.50e+000 |0000005.50 |
         %# |   555 | 01053 |   0x22b | 5.50e+000 |      5.50 |
         %0 |000555 |001053 |0000022b |05.50e+000 |0000005.50 |
          % |   555 |  1053 |     22b | 5.50e+000 |      5.50 |

    - Page 197 - The description of the scanf conversion type
      characters lists F to indicate a float conversion.  This
      is not correct.  F is not a valid type character.  It is
      used instead to override the default size of the corresponding
      address argument to be a far pointer.  That is, the format %F
      is not valid; it does not specify a float conversion.
      The format %Ff specifies a float conversion into an argument
      that is given with a far pointer.

    - Page 231 - The proper description for the strstr function
      is: strstr scans str1 for the first occurrence of the substring 
      str2.

    - The Merge string option (O/C/Code generation/Merge duplicate
      strings or -d) is actually default OFF, not default ON.

    - The discussion of text mode I/O in the functions close (p. 54) 
      and read (p. 188) states that Ctrl-Z is written to the end of a 
      text mode file when that file is closed, if that file was opened
      for write. That is incorrect. No Ctrl-Z is written for you. 
      If you want your text file to be terminated with a Ctrl-Z,
      you need to explicitly output one yourself. Some older programs on
      DOS require a Ctrl-Z at the end of text files. This is an outmoded
      practice inherited from CP/M.

    - The description of the intdos, intdosx, int86 and int86x
      functions is incomplete. The union REGS structure used by all
      four functions has the additional field x.flags. This field is
      set to the 8086 flags register after the system call is made.  
      This is in contrast to the x.cflag field, which simply contains 
      the state of the carry flag.

      Further, for int86x and intdosx, the segregs->es and
      segregs->ds fields are set to the values of the corresponding
      segment registers after the system call.

    - The variable int _doserrno is declared in dos.h and errno.h.

    _stklen

    Name         _stklen
    Usage        extern unsigned _stklen;
    Description

      _stklen is used to specify the size of the stack.

      In large data models (compact, large and huge), _stklen is
      the exact stack size in bytes.  In small data models
      (tiny, small and medium), _stklen is used by the startup code
      to compute the minimum size of the DATA segment:
           min DATA segment size =  size of _DATA segment +
                                    size of _BSS  segment +
                                    _stklen + MINSTACK (128 words).
      If the memory available is less than this, the startup
      aborts the program.

    _8087

    Name         _8087
    Usage        extern int _8087;
    Description

      The _8087 variable is set to 1 if an 8087 or 80287 is detected
      or if the 87 environment variable is set to Y (set 87=Y).  It is
      set to 0 otherwise.  You must have floating point code in your
      program for _8087 to be set to 1.


    3. IMPORTANT TIPS
    ÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄ

    - When copying the *.H files from the Libraries and Header Files
      disk be sure to copy the SYS sub-directory and its contents.

    - The C startup routine (c0?.obj) parses the command line
      to set up the argc, argv arguments to main. Double quotes on
      the command line can be used to bracket a single argument that
      has embedded blanks. For example, in the following, the
      program MYPROG.EXE is invoked with three arguments:

        myprog arg1 "arg2 with embedded blanks" arg3

      The second argument contains embedded blanks.

    - Here is further explanation of the switch to disable
      register variables (-r- or O/C/Optimization/Use register
      variables..Off).  Not only will the compiler not use register
      variables, but it also will not respect and preserve register
      variables (si,di) from any caller. For that reason, you should
      not have code that uses register variables call code which has
      been compiled -r-.


    4. FILES ON THE DISK
    ÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄ

    Disk 1
    ÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄ

    README   COM  --  Reads this README
    TC       EXE  --  Turbo C Compiler
    TCINST   COM  --  Installation program for TC.EXE
    TCHELP   TCH  --  Help file for Turbo C
    HELLO    C    --  Example Turbo C program
    README        --  This file!


    Disk 2
    ÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄ

    TCC      EXE  --  Command line version of Turbo C Compiler
    CPP      EXE  --  Turbo C preprocessor
    MAKE     EXE  --  Program for managing projects
    TLINK    EXE  --  Borland Turbo Linker
    TOUCH    COM  --  Program that updates a file's date and time
    CNVTCFG  EXE  --  Program to convert configuration files

    BUILD-C0 BAT  --  Batch file for building the startup code modules
    C0       ASM  --  Assembler source for startup code
    RULES    ASI  --  Assembler include file for interfacing with Turbo C
    SETENVP  ASM  --  Assembler source code for preparing the environment
    SETARGV  ASM  --  Assembler source code for parsing the command line
    MAIN     C    --  Alternative, stripped-down C main file

    Disk 3
    ÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄ

    ???????? H    --  Turbo C header files
    C0T      OBJ  --  Tiny model startup code
    C0S      OBJ  --  Small model startup code
    MATHS    LIB  --  Small model math library
    CS       LIB  --  Small model run-time library
    C0L      OBJ  --  Large model startup code
    MATHL    LIB  --  Large model math library
    CL       LIB  --  Large model run-time library
    EMU      LIB  --  8087 emulator library
    FP87     LIB  --  8087 library

    MCALC    C    --  MicroCalc main program source code
    MCALC    H    --  The header file for MicroCalc
    MCALC    PRJ  --  The MicroCalc project file
    MCALC    DOC  --  Documentation for MicroCalc
    MCOMMAND C    --  MicroCalc commands source code
    MCDISPLY C    --  MicroCalc screen display source code
    MCINPUT  C    --  MicroCalc input routines source code
    MCPARSER C    --  MicroCalc input parser source code
    MCUTIL   C    --  MicroCalc utilities source code
    MCMVSMEM C    --  MicroCalc direct screen memory write source code
    MCMVSMEM OBJ  --  Compiled version of MCMVSMEM.C - will link with any
                      memory model

    SYS DIRECTORY
    ÄÄÄÄÄÄÄÄÄÄÄÄÄ
    STAT     H    --  Turbo C header file with file status/directory
                      functions


    Disk 4
    ÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄ

    C0C      OBJ  --  Compact model startup code
    MATHC    LIB  --  Compact model math library
    CC       LIB  --  Compact model run-time library
    C0M      OBJ  --  Medium model startup code
    MATHM    LIB  --  Medium model math library
    CM       LIB  --  Medium model run-time library
    C0H      OBJ  --  Huge model startup code
    MATHH    LIB  --  Huge model math library
    CH       LIB  --  Huge model run-time library

    MATHERR  C    --  Source code for handling math library exceptions
    FILECOMP C    --  Example Turbo C program to compare files
    GETOPT   C    --  Parses options in commmand line

    CPINIT   OBJ  --  Initialization code to be used when linking
                      Turbo C and Turbo Prolog.
    BAR      C    --  Example function to be used with PBAR.PRO
    PBAR     PRO  --  Example Turbo Prolog program to be used with BAR.C


    5. USING TURBO C WITH DOS 3.2 AND A FLOATING POINT COPROCESSOR
    ÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄ

    DOS 3.2 has a bug in its handling of floating exceptions.  When
    an exception occurs, the system allocates a local stack for the
    exception handler.  Unfortunately, when a system stack gets
    allocated, it never gets deallocated.  The default number of
    stacks is 8.  After 9 floating point exceptions occur, the system
    hangs and a cold reboot is required.  Typical exceptions are
    overflow and divide by zero.  The stacks are not reclaimed when
    a program terminates, so a program which divides by zero once
    will crash the machine if run 9 times.  To avoid this problem,
    you can either:

        1. Not use DOS 3.2.
        2. Obtain a patch from IBM or Microsoft which fixes the
           bug, and install it on your copy of DOS.
        3. Not use a floating point coprocessor.
        4. Disable the exception trapping in your Turbo C program.
           You can do this easily by calling the _clear87() library
           function.  For example,

           #include <float.h>
           unsigned int fpstatus;
           /* stdlib.h included for exit, stdio.h for puts */
           #include <stdlib.h>
           #include <stdio.h>

           main()
           {
               _clear87();
               /* do floating point work here */
               /*  ...  */
               /* now, check for masked exceptions before exiting */
               fpstatus = _status87();
               if (fpstatus & SW_INVALID)
               puts("Floating point error: invalid operation.");
               if (fpstatus & SW_ZERODIVIDE)
                   puts("Floating point error: zero divide.");
               if (fpstatus & SW_OVERFLOW)
                  puts("Floating point error: overflow.");
               exit(fpstatus & (SW_INVALID | SW_ZERODIVIDE | SW_OVERFLOW));
           }


    6. HOW TO COMPILE MICROCALC
    ÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄ

       In order to compile MicroCalc do the following:

       With TC.EXE:

         1. Run TC.EXE
         2. In the Project pulldown menu specify the project name
            "MCALC.PRJ"
         3. From the main menu select the Run option

       With TCC.EXE:

         Compile from DOS with the following command line:
           TCC mcalc mcparser mcdisply mcinput mcommand mcutil mcmvsmem.obj

       In both cases, compiling under a large data model (COMPACT, LARGE,
       or HUGE) will give you much more memory for your spreadsheets.

END
ÄÄÄ

>