Magnum One

home *** CD-ROM | disk | FTP | other *** search

/ Magnum One / Magnum One (Mid-American Digital) (Disc Manufacturing).iso / d12 / mflms101.arc / TEKNOTES.DOC < prev next >

Wrap

Text File | 1989-11-25 | 9.8 KB | 397 lines

TITLE_PG.TN - Thu 23 Nov 89 00:20 - Page 1: The MicroFirm Function Library for Microsoft C & Quick C ver 1.01 Copyright 1988-89 by Robert B. Stout dba MicroFirm portions Copyright 1986-87 by Stephen E. Margison All rights reserved ------------------------------- TECHNICAL NOTES ON RELEASE 1.01 ------------------------------- TOC.TN - Thu 23 Nov 89 00:20 - Page 2: The MicroFirm Funtion Library - Technical notes on release 1.01 Page Topic 3 Object-oriented features of ver 1.01 - future C++ directions 4 High-level stream I/O 5 BASIC-like string functions 6 Julian (scalar) date functions OOPS.TXT - Thu 23 Nov 89 00:20 - Page 3: The MicroFirm Function Library includes several groups of functions which, not accidentally, lend themselves to an object-oriented view of programming. Future versions of The MicroFirm Function Library will, in fact, include C++ class implementations of some of these: Stream I/O ---------- The high-level stream I/O functions and their associated installable stream filter functions are the most obvious candidates for conversion to classes. These are already written to hide the functional code and data structures as much as possible from the implementation. BASIC-like string handling -------------------------- It seems everyone who does C++ classes includes a string class, so why should I be different? As such classes go, the functions in The MicroFirm Function Library are quite conservative, yet addresss the requirements of most programmers who are simply looking for the ease of BASIC-like string processing in C. Math ---- The specialized math routines for performing CRC calculations, integer square roots, and simultaneous divide and modulus can all be coded in C++ so that separate versions need not be called for different data types as is the case with the C versions. Directory operations -------------------- Although written for compatibility with BSD 4.3 the opendir(), readdir() and closedir() functions could well be treated as a C++ class. Filename processing ------------------- The MicroFirm Function Library's extensive facilities for normalizing filenames, explicit smart wildcard matching, etc. could be enhanced by the implementation of a DOS filename class. STREAMIO.TXT - Thu 23 Nov 89 00:20 - Page 4: The MicroFirm Function Library's stream I/O is a unique (for now) approach to high-level file I/O. Analogous to a shell's (e.g. DOS's COMMAND.COM) ability to pipe and/or redirect file I/O, the stream I/O functions provide the ability to install filters in the stream. Examples included in the release of The MicroFirm Function Library include filters to force streams to either upper or lower case, to encrypt or decrypt streams, and to perform simple run-length encoding. Note that the use of these filters is totally transparent to the application. Another unique feature of The MicroFirm Function Library's stream I/O is the ability to treat external data channels (e.g. interrupt-driven serial I/O functions) as if they are simply another stream device like a disk file. In order for this to work, the minimum requirement is for appropriate functions to exist which function analogously to the normal fgetc() and fputc() functions. A function to unget a character may also be required (see below). Filters installed in streams are installed as if on a stack and invoked in a last installed, first called fashion. Only one filter may be installed which alters the number of characters transmitted to or from a stream. An example of such filters are the run-length encoding filters included in this library. These are not particularly useful as written (sorry - ran out of time to do LZHuff or arithmetic coding filters), but demonstrate how to write such filters. Since such filters must have the ability to gobble or source characters not in the original stream, they must be stacked immediately before or after the stream destination or source, respectively. This means such filters acting on a stream opened for writing must be installed first in the chain (last called). Such filters installed in a stream opened for reading must be installed last (first called). Note that, since character sinking or sourcing is done within such a filter, the character count reported by the sfread() and sfwrite() functions is the number of characters without such processing. For example, if I read in 1024 characters from a file then pass them through a compression filter to another file, sfwrite() will return 1024 as the number of characters written even though the number of characters physically written was smaller due to the action of the compression filter. Similarly, if I read the compressed file back through an expansion filter, sfread() will report that I read 1024 character from a file which may have been physically smaller since it was compressed. There are two other requirements of filters which must have the ability alter the stream's character count. If such a stream is opened for reading, it must provide an unget-equivalent function to allow it to replace unused characters in the stream. If a stream is opened for writing, it must provide a flush function to force unprocessed output to the stream when the stream is closed. BASTRNGS.TXT - Thu 23 Nov 89 00:20 - Page 5: The MicroFirm Function Library provides the unique ability for C programmers to process character arrays (strings) as simply as most BASIC programmers (or C++ programmers!) It does this by maintaining and reusing a pool of string space. Note that since these strings are reused, to retain permanent copies requires the use of strcpy() or strdup(). This is an inconvenience at times but avoids the overhead inherent in "garbage collection" algorithms such as used by BASIC. Most times, the strings will be used before their string space is reused. The primary function of this package is stralloc() which returns a string buffer from the pool which is guaranteed to be large enough to hold a string of the requested size. The stralloc() function is called internally by several of the other functions and is available for use by applications needing a temporary string storage buffer. The string pool defaults to 16 strings in small and compact models and 64 strings in large and medium models, with all strings dynamically resized as required. Where these are too few, the str_init() function allows up to 128 strings in small and compact model or up to 512 strings in large and medium model. To avoid possible problems when used by recursive functions or functions otherwise using malloc(), the size of the strings maintained may also be set to a constant via str_init(). The str_free() function frees the entire string pool. In addition to left(), right() and mid() corresponding to BASIC's LEFT$, RIGHT$, and MID$ functions, The MicroFirm Function Library adds the string_add() string concatenation function as an analogue of BASIC's "A$=A$+B$+C$" syntax. JULFUNCS.TXT - Thu 23 Nov 89 00:20 - Page 6: The Julian date conversion routines convert dates in standard year, month, and date formats to and from Julian dates. Note that these are not true Julian dates based on the Julian calendar, but are scalar numbers representing the number of days since Jan 1, 0001. As such they may be used for calculating the number of days between two dates, etc. Two additional functions are supplied for converting Julian date to day of the week and day of the year value. The first day of the week conversion function, julian_to_wkday(), returns the numerical day of the week, starting with Sunday equal to zero. The second day of the week conversion routine, julian_to_dayname() returns the name of the day as specified for the current locale - see setlocale() for more details. In addition to the basic Julian date routines, four symmetrical functions are provided to convert Julian dates from and to both time_t values and/or tm structures. EXAMPLE unsigned mo,day,yr; long jday1,jday2; jday1=ymd_to_julian(1903,1,3); jday2=ymd_to_julian(1974,6,9); printf("There were %ld days between 1/3/1903 and 6/9/1974\n", jday2-jday1); printf("6/9/74 (day # %d) was a %s\n", julian_to_yrday(jday2), julian_to_wkday(jday2)); printf("That is %f years\n", (double)(jday2-jday1)/365.2425);