Nash - Scientific Computing with PCs, страница 13

In our experience, simple text files, usingprintable characters of the ASCII 7-bit standard set, are the only files that can be transferred and used withany assurance between systems. Word processing files and binary data files are notoriously awkward totransfer (Nash J C, 1993a).The access mechanism to files is determined by the computer we are using, its operating system and thesoftware used to create the files. The user, however, generally has a choice in the naming of files.

The goalshould be to ensure that names chosen are meaningful within the context of the work or subdirectory athand and within the name length and character restrictions imposed by operating or application software.Note that directories and subdirectories are themselves files and should also be appropriately named.The MS-DOS file system has been criticized for having many shortcomings. One of these is that thefilenames may not exceed 8 characters in the main part of the name and 3 characters in a filename32Copyright © 1984, 1994 J C & M M NashNash Information Services Inc., 1975 Bel Air Drive, Ottawa, ON K2C 0X1 CanadaSCIENTIFIC COMPUTING WITH PCsCopy for:Dr.

Dobb’s Journalextension. Typically the two parts are separated by a period (.) though other ways of writing the two partsexist. The blank character is not allowed in names although it is sometimes possible by error to get a blankembedded in a name. This generally leads to some awkward work to correct the name, or even to deletethe file.The filename extension is often used to suggest the function of a file. Indeed, the MS-DOS operatingsystem treats files with the extensions .EXE or .COM as executable. Files ending in .BAT are sequencesof operating system commands and may also be "executed". Naming non-executable files with suchextensions can lead to unexpected results.

For example, it may be tempting to name a file of commentson a lecture as LECTURE.COM. Later, forgetting that the file is a text or a word processing file, we maytry to execute it.Operating systems may permit characters other than letters or digits in filenames. For example, MS-DOSallows us to use$%’-@{}~‘!#()&_The last of these, the underscore (_) is quite useful if we would like a "space" in the name e.g.MY_FILE.TXT. However, take care as some application programs may not allow the use of all of the abovecharacters.

To avoid errors, it is therefore advisable to use only letters and numbers. Foreign or graphicscharacters in filenames or extensions may cause some software to react in unpredictable ways.Filenames in MS-DOS are handled as if in upper case letters, but they can be entered with any mix orupper or lower case. That is, the filenames are case insensitive for MS-DOS. UNIX is case sensitive.For MS-DOS there are several characters that should not to be used in file names or extensions, namely?. , : ; = * \/ + " > <Indeed, the ? and * characters are wildcards: the ? takes the place of a single character, the * takes theplace of any character string starting at the location of the asterisk in the filename or extension.The Macintosh FINDER allows for longer filenames — 31 characters as the maximum.

The names can beupper or lower case and the name displayed stays in the case in which it was entered. However, wecannot create two separate files AAA and aaa. That is, the case sensitivity extends only to the display ofnames, not to search or use. The long names may include spaces and special characters. Thus theMacintosh names can include quite a large amount of identifying information. Unfortunately, transfer ofthe files to other systems may force us to use a more restricted set of names.The choice of names and extensions is, for the most part, a personal one.

We can nevertheless suggestsome we have found useful. Above, we have mentioned that MS-DOS (and other operating systems) mayuse the extension as an indication of the file usage. Besides .COM, .EXE and .BAT we have.SYS.OVL.BAS.OBJ.FOR.PAS.C.Hsystem filesoverlay (for an executable program)BASIC source codeobject code (input for the LINKer)FORTRAN source code (though .FTN also used)PASCAL source codeC source codeC header fileSeveral applications create specially named files. Unfortunately some of the names are conflicting oroverlapping and may include:5: FILE MANAGEMENT.DTP.DOC.TXT.WK?.TIF.ARC.LZH33DeskTop Publishing filedocument file (used by several word processing programs and also generically to indicateDOCumentation)text file, also used by some database programs for databases stored in text formworksheet for various spreadsheet programsgraphics file (Tagged Image Format) used (in different ways!) by various drawing,photographic and fax programsan archive file, not always in the form used by programs such as ARC (Systems EnhancementAssociates)an archive file created by LHA (Haruyasu Yoshizaki, 1988)In our own business work we have used extensions such as the following to keep track of common typesof files:.BID.INV.LET.LST.MTG.PRO.REC.RPTbidinvoiceletterlistmeetingproposalreceiptreportWhen working with files that have specific temporal importance, we can also think of using abbreviatednames for months, e.g., .JAN, .FEB, .MAR, etc.

or for days of the week, e.g., .MON, .TUE, .WED, etc.There are as many proposals for good file extensions as there are people using computers. Since summarydirectories do not show file time/date stamps and some software automatically updates this informationwhen used to view the file, it is often helpful to include some temporal information in the file name, e.g.,NASHJC.92B for the second letter sent to J.C.Nash in 1992. On the negative side, we may forget howmany letters we have written this year to active correspondents.Returning to the filename rather than the extension, we note that there may be reserved names. MS-DOSreserves AUX, COM1, COM2, COM3, and COM4 for serial communication ports, PRN, LPT1, LPT2, andLPT3 for printer ports, CON for console input (keyboard) and output (screen), and NUL as a dummydevice to which output can be directed.

These predefined "devices" are always available to programs. Forexample,COPY A:*.* NULchecks that all files on A: are readable. Unfortunately, documentation of these device names is scatteredthrough various parts of the MS-DOS manuals and books about this operating system. Worse, in a recentattempt to move files from a UNIX system to a PC, a series of files related to "conversion" had filenameCON with different extensions. As they were downloaded by modem, the communications programcheerfully copied them to the screen rather than the disk!5.2Types and Distributions of FilesTo give meaning to the discussion, we present here a brief analysis of the files found on just one of ourIBM-class PCs. The purpose of this is to present a snapshot of a file collection, indicating the number, sizedistribution and general usage of files in a machine.

We want to control the files — to make sure we donot lose information but are not overwhelmed by too many copies and versions of each file. Clearly thecharacteristics of a set of files will depend heavily on interests and inclinations of the user(s), but the ideasare general.At the time the snapshot of all files was taken, there were 2474 files or directories in the fixed disk. Of34Copyright © 1984, 1994 J C & M M NashNash Information Services Inc., 1975 Bel Air Drive, Ottawa, ON K2C 0X1 CanadaSCIENTIFIC COMPUTING WITH PCsCopy for:Dr. Dobb’s Journalthese 1119, or about 45%, we could identify as system or software files, while 1355 (about 55%) were filescreated by the user(s).

This is a very high ratio of operating and support files to application files. The ratiocan be expected to be lower in a system that is mostly dedicated to particular applications such as wordprocessing. Our uses, involving testing and evaluating software, weigh the balance in favor of system andsupport files.Some of the files on disk were duplicates; of these, few are intentional copies, such as safety orconvenience copies of configuration files or style sheets that save us changing directories. We also notedsome "temporary" files of various types as well as a few files of unknown origin.

These should, of course,be deleted. The distribution of filename extensions is highly skewed, as shown below: some extensionsare heavily used, but many appear once, suggesting our discipline in naming files is imperfect.Figure 5.2.15.3Distribution of filename extensions in sample set.Cataloging and Organizing FilesWhenever files are stored, we would like to know and possibly have a list of the following information:•File name;•Location of primary and backup copies;•Date and time of creation and/or modification;•Length of file;•Format or type of file, i.e., how it is structured;•Description of contentsWe know of no operating system for which such a complete list is automatically kept. Nevertheless, thesix points above provide a set of desiderata to which we may aspire.In what follows, we will be referring to our entire collection of files, both on-line and stored on externalmedia.

In practice, unless we are very assiduous, we will not even have a record of file names andlocations. While most PC operating systems include file lengths, there may be no information about thefile format and the time/date stamp of the file is only meaningful if the computer that created it had acorrectly set clock/calendar. Moreover, users are often careless in file manipulations — there is a strongtemptation when "looking at" a file with an editing program to automatically save it again, therebyupdating the time stamp. Some software is perverse enough to do this for us!We strongly recommend that the date and time be checked every time a machine is started.