Nash - Scientific Computing with PCs, страница 5

There are a bewildering number of features and choices to make, as well as several interfacemechanisms to consider. Of these, we note in particular TEX (Knuth, 1986), which is perhaps the mostimportant of the mark-up languages. In a mark-up language, all the material is entered as text in astandard alphabet with special commands embedded to indicate fonts, positioning, type styles, specialsymbols, underlining, etc. The commands use characters from the standard alphabet.

TEX preparesmaterial for type-setting, but the source material is not particularly readable to the uninitiated. Thealternative is a WYSIWYG (What You See Is What You Get) approach where the user selects and positionssymbols, possibly with the aid of some built-in formatting rules. An extensive surveys of scientific textprocessing has been carried out under the direction of Dr. Richard Goldstein (Goldstein et al., 1987a,1987b, 1987c).As special-purpose scientific text processors have evolved, the general word-processing vendors have alsosought to add scientific features to their products. This book has been prepared using WordPerfect 5.1(abbreviated WP5.1 from now on) using an MS-DOS computer with a VGA screen.

We chose to use WP5.1mainly for reasons of safety rather than features or convenience. In particular, with the very large installedbase of users, there was likely to be advice and assistance available to overcome difficulties and userUNfriendly quirks, of which we encountered many.2.6Macro-editors and Data ConvertersScientific computation inevitably requires data to be edited or its format transformed so that programscan make use of it. Lack of appropriate tools for these tasks is perhaps the most common source offrustration for researchers attempting to use PCs in their work. Sometimes editing and reformatting canbe accomplished with a suitable macro-editor as discussed below.Most PC users have some form of text editor.

Frequently it is the same editor that is used for preparingletters or reports. Unfortunately, there are some significant dangers for the unwary. Students who havelearned a word-processor such as WP5.1 frequently do not realize that the "document" files saved byword-processors include a large amount of information on how the document is to appear on the page.This information is stored as bytes in the file additional to the text in which we are interested. When anapplication program such as a programming language compiler reads the file, anticipating only the text,errors occur.Even as a tool for looking at results, word processors have deficiencies. For example, it is quite simple tobuild a spreadsheet of fifteen columns of numbers. Allowing 10 spaces per column, this is 150 spaceswide.

If we import the "printout" from such a spreadsheet into a word processor, we may find that thepleasant tabular structure of columns and rows has become an almost unreadable jumble of numbers andcolumn headings. The difficulties are:•Word processors often allow users to set and change margins and font sizes.

Text is then wrapped tofit the margins. That is, the line lengths are adjusted to fit the margins, not vice-versa. We need to setour margins wide enough to accommodate the 150 character-wide lines of our spreadsheet if we wishto be able to see the structure.•Word processors usually allow proportionally-spaced fonts. In these, the number 1 may take less spacethan a number 9 or a 7.

The column alignment is once again lost unless we specify that a monospacedfont is to be used.This does not imply that one cannot use a familiar word-processing editor, but care needs to be taken.It is quite easy to find public domain or freeware editors that allow for the manipulation of the text alone.(NED — the Nash EDitor — is our own freeware contribution.

At under 6K bytes, it is very small andfast, yet includes a help screen and mouse support. See Nash, 1990a.) Users may also find that there areconvenient editing tools included with programming languages or even with the operating system.Microsoft, after many years of imposing the truly medieval EDLIN on MS-DOS users, incorporated the10Copyright © 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 JournalQuick BASIC editor EDIT, in MS-DOS 5.0. Such tools serve very well to correct or adjust a few charactersin a file but are less useful when many changes have to be made.For manipulations that have to be repeated, we recommend a macro-editor, in which the user can set upseveral sequences of commands that can be executed via a single keystroke.

This vastly simplifiesoperations of removing unwanted commas in data files, stripping out or adding in spaces, and similartedious but necessary operations.When we have a large (>50K) file to manipulate we use the commercial version of NED.

It can edit anyfile for which we have enough disk space. One common use of such editors is to search through large textfiles for information.We find that we use variations on the above theme nearly every day. Clearly we could use a special textsearch utility like the UNIX facility GREP. We occasionally use file viewers that allow searching, butprefer a small set of very familiar working tools. Moreover, the editor allows us to buffer and extractpieces of the text in which we are interested.Clearly a text editor cannot easily adjust a file that is encoded in some way. For example, spreadsheet datafiles contain a variety of different forms of data: numbers, strings, data ranges, printing and output controlinformation, and graphs.

The typical worksheet is saved in binary form and is not amenable to viewingwithout special programs. Similar comments may apply to the files saved by database managementprograms, though some packages advertise the important fact that information is stored in text files.We usually find it easiest to extract information from spreadsheets or database programs by "printing"information to a text file and editing it with our text editor. We use the spreadsheet or database programto select which data we save to the file, as well as the order in which it is presented, while the text editoris used to adjust the detailed layout or format. Here we are following a strategy of choosing to usefamiliar tools and a tactic of using each to perform tasks for which it is best suited.Alternate approaches exist.

In circumstances dependent on the application and frequency of use, we mayprefer to use the data translation facilities of programs we use. Spreadsheet, database, word-processing,and statistical programs frequently provide selected translation facilities, especially into, but lesscommonly out of, their native formats. We note, for example, the Translate facility of Lotus 1-2-3. Whenbuying software, it is important to note which data formats can be used.More generally, there exist programs that attempt to convert a range of different data formats, even acrossoperating systems or computing platforms. Usage will depend on application.

For example, MACLINKconverts file and disk formats for a variety of word processing programs between the Apple Macintoshand MS-DOS families of computers. DBMSCOPY allows data exchange between a large collection ofdifferent database management and statistical programs. Graphics Workshop allows a variety of picturefiles to be transformed one into the other, mostly successfully, but always with the danger of some lossof definition or "out of memory" error. CorelDRAW and CorelPHOTO-PAINT have proved more flexible.Unfortunately, data file formats are constantly evolving, so that the provision of "long-lasting relief" tothe data format transformation problem is not likely to be found. Our strategy has been to try to save dataas much as possible in text form, since users then have a reasonable chance to find some way to workwith files we provide.

In this regard, we note that both TEX and PostScript, which essentially describegraphical information, are coded as text. PostScript files are rather difficult for humans to interpret, butthere are many different machines — usually laser printers — that can do so. TEX, however, we find easyto convert to simple text or to camera-ready material, even though we do not have the software toperform the manipulations. The main codes are simple enough to follow so that we can provide theequivalent word-processing commands.2: DATA PROCESSING CAPABILITIES OF PCs2.711GraphicsSome computer peripheral devices literally "draw" lines on either paper or a screen by moving a pen oran electron beam from one point to another, then another, and so on.

For this reason, they are referredto as vector graphic devices. The TV screen and most computer monitors draw dots on a screen in a rowof lines. To draw a picture, one must put the appropriate colored dots in the correct position in a largematrix of dots. They are called raster graphic devices. Working on the same principle, drawings can beoutput to dot-matrix printers, of which the laser printers are a high-resolution, small-dot version.It is not trivial to use raster devices directly, since we think of drawing, e.g., with a pencil, in vector form.The "programming" of drawing is easiest with vector graphic tools, for example, the Hewlett-PackardGraphics Language (HPGL) that can drive not only plotters but also Laser printers and even numericallycontrolled machine tools.

PostScript may also be considered a vector graphics language.For research users, there are likely to be many reasons why graphical computing tools will be important:•Research data is frequently best presented as a simple graph, for example, growth of plants or animalsover time, chemical reaction kinetics, or spacecraft trajectories. Statistical distributions and datarelationships can be visualized by a variety of data plots.•Drawings of instruments, either in preparation for their construction, to illustrate what we proposeto do, or to report what has been done, are now more easily prepared with Computer Aided Design(CAD) software than by recourse to a draftsman.•In a technical report, we commonly need to include a photo.