Chapter 15
Contents
1. tion the programmer simply bypasses the old instructions and adds new duplicating whole sections Modular programming is by definition concise When modules are main tained it is easy to remove a task and rework the solution as a new module or function Good programs should be efficient Efficiency is usually measured in terms of the time taken for a task to complete Many simple measures can be taken to make programs efficient some of which can make vast improvements in execution time at very little cost Good programs should be usable Programs that are not usable probably have only one user the author Worse than this once the author is finished using the program it will be forgotten Modular programs encourage usability because modules can be interchanged A programmer using structured programming rarely starts a new program from scratch General purpose functions can be used and reused in a variety of different programs and contexts Good programs are documented and maintained Documentation it is arguable is more important than the program Documentation can be internal in the form of headers and comments or external in the form of written text either interactive such as help screens or paged External documentation should be able to function in a user or tutorial capacity for the new user and as reference for the experienced user Often this means two manuals a user manual and a reference manual Good documentation is understandable2. it can be read by a novice user and understood Good software is maintained Good software has somewhere near the beginning a ver sion number and the version number is high and decimal such as 4 03 Constant updates mean that a programmer preferably the author has gone to the trouble of finding out what is wrong with the software and has changed it to make it better Usually for small changes the decimal place is increased For major revisions of a program the version number is increased Finally good programs work This means that they produce the correct answer which is not always the answer we expect or that we want Working programs must be reliable in two ways First they should survive changes in the operating system Second pro grams should be internally consistent For example a statistical program should give the same results using the same data more than once This implies reliability in terms of con sistency and in terms of resilience to change We have already noted the special demands of cartographic and geographic data The volume of data that need to be handled influence how we write programs that work with geographic data Geographic programs also have to run on a variety of different types of computers because no one computer is best for every geographic problem Cartographic computer programs very often use graphic output rather than sets of sta tistics or numbers Cartographic programs use geographic data structures ins
3. of computer it is and how to start the boot or the initialization process At the next level is the machine language program This language has to be tailored exclusively for a particular brand of computer a particular type of memory and so forth Programs in machine code language talk directly to the hardware Machine code has the disadvantage that not only are data required to be in binary but the instructions are also The last of the low level computer languages is assembly language Assembler is com piled in the same way as higher level languages This means that a program can be en tered and stored in a file then processed into machine language via a large translation program called a compiler The machine language program is then ready to run An assembler gives access to registers or parts of the memory into which we can write instructions and numbers Assembly language uses as building blocks an instruction set consisting of multicharacter mnemonics standing for individual operations These oper ations codes perform simple operations such as putting a number into memory or taking the current number that is in one register and performing a binary AND with the number in the current register Assembler language programs are highly adapted to the architec ture of the computer on which they run and as such are capable of running very fast In many cases programs are written in higher level languages and their most time consum ing modul
4. 15 1 15 Cartographic Computer Programming LANGUAGES The control necessary for implementing cartographic data structures for performing car tographic transformations and for anything other than applied computer cartography re quires the use of a programming language Many students find their way into analytical cartography from other disciplines and as such are excluded from a more in depth under standing of analytical cartography because they do not know how to program Although much of the material presented in this book can be understood and even taught without the use of programming the advanced student may wish to go further If your background has not been in computing the best advice is to learn programming from the experts This usually means in a computer science department Learning programming along with a graphics course is not recommended When we program we usually do so in a programming language Just as we use the English language to express our ideas to others so we use a programming language to express our ideas to the computer Computer programming languages perform sequences of instructions This is the case even though our thoughts are often not sequential A pro gram moves from step A to step B to step C and the move to step C does not start until B is reached Human thought can easily branch from A to C or to X Y and Z or all four simultaneously based on the loosest of connections Usually a program o
5. and software change flavor with the times If the early days of computing can be classified as the era of hardware then the cur rent generation will clearly be the era of software and above all programming standards Links with the standard version of C now approved at the ANSI level and with the Spa tial Data Transfer Standards ensure that the duplication of effort and the time wasting of the early days of analytical and computer cartography are over and that cartographers can concentrate upon the more substantive aspects of their discipline safe in their knowledge of how to produce the map 15 4 REFERENCES Enderle G K Kansy and G Pfaff 1984 Computer Graphics Programming GKS The Graphics Standard Symbolic Computation New York Springer Verlag Hearn D and M P Baker 1986 Computer Graphics Englewood Cliffs Prentice Hall International Standards Organization 1985 Information Processing Systems Comput er Graphics Graphical Kernel System GKS Functional Description ISO Pub lication No 7942 1985 Kernighan B W and D M Richie 1978 The C Programming Language Prentice Hall Software Series Englewood Cliffs Prentice Hall Kernighan B W and D M Richie 1988 The C Programming Language 2d ed Pren tice Hall Software Series Englewood Cliffs N J Prentice Hall Moellering H 1983 Designing Interactive Cartographic Systems Using the Concepts of Real and Virtual Maps Proceedi
6. dards Two major graphics standards have found their way into analytical and computer cartography The first of these CORE was an out growth of a workshop group of SIGGRAPH the Association for Computing Machinery Special Interest Group on Graphics in April 1974 The early standard was published in 1977 with completion in 1979 This standard was important in establishing terminology and concepts The standard was broadly implemented mostly with FORTRAN bindings during the 1970s and 1980s Starting in 1979 an ANSI Technical Committee on Com puter Graphics used the CORE system as a model in designing a now well established two dimensional standard called GKS Almost all the facilities available in CORE were carried over into GKS and its three dimensional equivalent GKS 3D CORE showed users the advantages of separating modeling and viewing functions 300 Cartographic Computer ProgrammingChap 15 This means that a graphical object such as a cartographic object can be defined given attributes and so forth independently of the geometry and properties of the final image an idea termed virtual maps Moellering 1983 CORE s weaknesses were the concen tration on line drawing functions its inability to support a multiuser workstation envi ronment and the failure to carry the standard over to include language bindings This last weakness meant for example that function and subroutine names the order of arguments in function call
7. e are LISP PROLOG MODULA SNOBOL and the object oriented languages such as C and Smalltalk Beyond the high level language are application level programs or macros within systems with their own internal programming language Graphics languages exist that can perform manipulation of graphic objects and data structures with a command lan guage rather than a programming language For example within AUTOCAD you can program with LISP or in many systems you can query a database using SQL Standard Query Language The language used throughout this book is a language called C C is the third version of a programming language developed at Bell Laboratories by Dennis Richie Originally written on a DEC PDP 11 for the development of the UNIX operating system the lan guage is general purpose and terse it supports advanced control and data structures and itis free of many of the restrictions placed on other languages by their operating environ ments The language was chosen for this book because C language programs are brief pow erful and support the complex data structures required for cartography and because C is highly portable to different computers and operating systems The language was original ly set forth in a book The C Programming Language by Brian Kernighan and Dennis Richie 1978 This book set the early C standard K and R C and is the standard fol lowed in this book Since then the American National Standards Institute has endo
8. ebuilt from data Different trans formations can be applied to whole segments Similarly higher level implementations of the standards support the metafile Sec 15 3 Software Development Methods 301 The metafile is a generic storage space that behaves the same as a device or worksta tion Thus to save the outline of an island a collection of polylines representing the island could be named as a segment be written into a metafile and then later during the same program run or even later by another computer program be retrieved and be redrawn with different transformations or attributes It is sometimes possible to retrieve these metafiles from different programs perhaps even in different languages An additional feature of the standards is the ability to cluster groups of attributes into bundles A bundle can be selected and uniformly applied to either segments or primitives For example we could define a set of attributes that should apply to map titles such as boldface fonts large letters a specific spacing of letters a particular color and text cen tering By referencing this bundle we can apply it to a specific piece of text that is to be a title rather than changing all the attributes as we go A typical computer graphics program sets the attributes after having computed where it wants to draw primitives opens a segment simultaneously defines and draws a seg ment and then closes the segment For example to draw a solid red po
9. es are rewritten in assembly language for efficiency The remaining levels of computer programming languages fall into what are called high level languages High level languages have named operations expressions that per form control and defined data types Examples of widely distributed higher level lan guages are FORTRAN or FORmula TRANslation language although it certainly was not the only one BASIC and COBOL Others have found their way into general use A distinguishing characteristic of these languages is that although they contain the mechan ics to write modular programs they do not make the modular structure an integral part of the language One step beyond high level languages like FORTRAN are the structured languages Structured languages like Pascal and Ada encourage the user to write programs that are modular Modular programs consist of independent units that can be worked on in isola tion from each other A structured language encourages the splitting up of programs into smaller and smaller pieces Each small program can be worked on tested and optimized independently and then the modules can be assembled into a whole working program 296 Cartographic Computer ProgrammingChap 15 Other structured languages are PL 1 ALGOL APL C and RATFOR In addition a number of special purpose languages that are suited to specific tasks such as text and list processing and complex database management exist at this level Among thes
10. gnificantly increasing the ability of software to execute in the same way under a number of different GUIs and operating systems 15 2 GOOD COMPUTER PROGRAMS Given a group of programmers with different levels of experience and a problem perhaps even a problem solved by a published algorithm it is most likely that each programmer will write a different program to solve the problem Some programs will not work in which case it is easy to tell whether or not the solutions have value Among the working programs however how can we tell which is the best or even which are good To be able to do so requires that we can tell a good program when we see one Fortunately good computer programs have some distinguishing characteristics so much so in fact that it is possible to have a good program that does not work and a bad program that does work In this section we will try to cover some of the factors that make computer programs good To begin with a good computer program is readable This means that programmers can read the program working their way through it without spending an undue amount of time trying to figure out what the original programmer meant Clean code is neatly formatted usually by a beautifier program and is well presented A good computer program is structured A structured program has embedded within itself subprograms that do smaller and smaller tasks so the main program consists of very general tasks This
11. ing follows from a familiarity with and experience in the C language Calls to GKS functions are as simple in C as in C because the primary differences relate to how data are modeled see Chapter 10 C allows the programmer to group together data and operations such as the actions produced by different functions together in a class The graphic object of a polygon en coded as a RING structure for example could have a set of functions that relate to it Sec 15 2 Good Computer Programs 297 area computation point in polygon testing or centroid specification among others The complete specification of a set of classes or objects of geographic interest remains to be done A subset of cartographic objects seems more feasible and is likely to follow the specifications of the DLG E descriptions advanced by the U S Geological Survey Finally C allows the specification of complex objects such as windows user dia logs and events This means that C is particularly useful when the programmer wishes to use the GUI toolbox directly Windows and X windows coupled with C for exam ple allow the programmer access to the building blocks of the user interface with which the operating system itself is constructed giving significant power of control to the pro grammer A consequence for example is the ability to inherit the characteristics of the operating system s windows by an application after compilation More extensive use of C is si
12. lygon first we need the boundaries of the polygon usually from data Next we assign characteristics such as a solid red fill With this done we simply open a segment draw the polygon and finish Once drawn and defined the red polygon can be recalled very easily just by referring to its segment identifier In Chapter 16 we will see how to write a graphics program using the GKS standard The GKS standard has become an important part of graphics programming especially since the standard s adoption by the American Standards Institute and by the Internation al Standards Organization in 1985 The standard includes a functional description of GKS ISO 7942 1985 the metafile ISO 8632 1 2 3 4 1987 and bindings for FORTRAN ISO 8651 1 1988 Pascal ISO 8651 2 1988 and Ada ISO 8651 3 1988 The three dimensional extension of the standard was approved by ISO in 1988 ISO 8805 1988 The ISO review of the standard involved over 100 scientists and about a 50 person year labor effort The standard has been adopted for several books in computer graphics including Enderle Kansy and Pfaff 1984 and Hearn and Baker 1986 Although other standards discussed in the following section are often more suitable for specific graph ics and even some cartographic applications GKS remains the benchmark against which all computer cartographic software in the future will be measured 15 3 2 Higher Level Graphics Languages and Standards Standards in the c
13. ngs AUTOCARTO 6 Sixth International Symposium on Computer Assisted Cartography Ottawa October 16 21 vol 2 pp 53 64 National Computer Graphics Association 1987 Standards in the Computer Graphics Industry Fairfax VA National Computer Graphics Association
14. omputer graphics industry have been slow to develop yet offer a large number of advantages to the programmer A standard expresses a nominal set of require ments a minimal level of performance and a mandatory area of compliance and con formance to accepted specifications NCGA 1987 In the United States it is the American National Standards Institute ANSI that re views and approves standards and represents the United States in the International Stan dards Organization The X3H3 committee of ANSI started examining standards for computer graphics in 1979 The X3H3 committee initiated GKS the Computer Graphics Metafile CGM and 302 Cartographic Computer ProgrammingChap 15 the Programmer s Hierarchical Interactive Graphics Standard PHIGS the Computer Graphics Interface CGI as well as GKS 3D GKS 3D extended GKS to handle the definition and viewing of three dimensional wire frame objects The system supports hidden line and hidden surface removal in the workstation transformation but does not support rendering techniques such as shading and light sources GKS 3D places a significantly enlarged set of demands upon the host computer particularly the amount of memory used and the processing power required As such microcomputer applications may be limited to only the most powerful machines The Computer Graphics Metafile is a file format suitable for saving a snapshot of graphics in their final form Access to images sto
15. otter a camera or a window of a workstation As we saw in Chapter 11 with a graphics standard we use three different coordinate systems We call the coordinate system of the application the world coordinate system Graphics standards use normalized device coordinates NDCs which are coordinates on a virtual device A virtual device is an idealized planar space on which the map is to be drawn Under GKS the lower left corner of the NDC space is at 0 0 and the upper right corner is at 1 1 Inthe CORE system which supports three dimensions direct ly the center of the screen is 0 0 0 the upper right corneris 1 1 0 and the lower left corner is 1 1 0 At the display end there are workstation or device coordinates which are locations on the display surface as referenced using the mechanics of the display system Under a graphics standard the specifics of the device coordinates are unnecessary although we can inquire their values if need be In addition once a window and viewport transforma tion have been established the programmer can reference all locations with map coordi nates and specific locations on the map such as locations of legends titles insets and so forth in normalized device coordinates Graphics standards also allow maps to be structured At the lowest level the map con sists of primitives Primitives can be collected into segments Segments can be named numbered and recalled by reference instead of being r
16. program passes program control to more and more specific tasks each of which works independently of the others and returns control when the subtask is complete This task nesting can make programs very manageable and can reduce programming time signif icantly Structuring within some languages such as within FORTRAN is by function and subroutine In the C language subroutines do not exist only functions So in C a program is defined as a function that is itself a set of functions The philosophy behind structuring is that programming complexity is worth minimiz ing Writing tiny programs that do very specific tasks is comparatively easy so if we can take a major task and divide it into many small tasks each of which can be solved by writing a very simple program programming complexity is minimized This approach could be called the divide and conquer approach because a difficult problem divided into less difficult problems and finally into easy problems expresses the original prob lem in a way that can be easily solved Good programs are structured and the structure is meaningful in the terms of the problem and works as a solution 298 Cartographic Computer ProgrammingChap 15 Good programs are concise meaning that they do not contain any more instructions than are necessary to get the task done A major cause of verbose programs is the re writing performed when program maintenance is done Rather than reworking the solu
17. r set of programs written in a programming language performs a specific task Such a program is referred to as an applications program The other general division of programming is systems programming in which the programs control the op erating system itself All cartographic programming falls into the applications program ming category although there is much that cartographers have to gain from an understanding of systems programming 294 Sec 15 1 Languages 295 All computer programs perform certain generic processes Most programs perform in put that is they read data from files or a user they perform operations that transform the data in some way such as between data structures and then they produce output The output can be a column of numbers a new file a graphic or a map The most important function of a computer programming language is to control the operation taking place The programming language gives complete control over an application it allows us to say exactly what we want to do how we want to do it what data should be processed and where we want to put the output Languages have different levels At the very lowest level of instruction computers use programs called microcode Microcode is in binary and is so fundamental to the opera tions of the computer that without it you cannot even load the operating system or use the computer Microcode instructions tell a computer where to find the operating system what kind
18. rdware Although very fast programs that are written in this way usually only work on one type of graphics configuration and on one specific computer Different computers have different screen sizes different colors different memory addresses and address ranges and different ways of mapping a byte onto bits and they support different numbers of col ors line thicknesses and so forth Not only must each graphics program be rewritten for each display device it must also be rewritten for each computer In a rapidly changing hardware environment to be device specific means almost instant obsolescence As a response many software manufacturers offer graphics toolkits which allow some degree of program portability but are still very device specific Many program ming languages such as microcomputer based versions of languages like C and Pascal offer graphics extensions or functions that produce graphics At the very simplest a MOVETO and a DRAW are provided which allow the user to move to a screen pixel location the geometry of which is the responsibility of the pro grammer with or without drawing a line from the current location Many toolkits sup port some sophisticated options such as graphics text circles ellipses and polygon fills but the programming level is still fairly close to the machine and the resulting programs while a little more portable are still highly device specific The alternative is to use graphics stan
19. red in a metafile is either sequential or random access and the image is defined in a device independent way The purpose of the CGM is to allow images generated on different graphics systems to be interchanged and regenerated on different computing and display environments The three parts of CGM contain the character encoding for assisting in network communications the binary en coding to support transfer between machines with different word sizes and operating sys tems and the clear text encoding to ensure maximum readability for use and debugging PHIGS like GKS is a functional definition of the links between a programming lan guage and a graphics system PHIGS supports some operations unavailable in GKS in cluding the ability to nest graphical objects hierarchically important in modeling applications and CADD and graphical object database support PHIGS is especially use ful when complex objects are to be constructed from smaller simpler objects for exam ple a fan can be constructed by defining a blade and then repeating it with rotation PHIGS supports three dimensions and interactive graphics such as panning and zooming at the workstation level The increase in PHIGS s flexibility and complexity however means that many workstations particularly at the lower end of the power scale are not capable of supporting all the capabilities of a full PHIGS implementation PHIGS has come to be used therefore in high end interactive works
20. rsed a version of C ANSI C The second edition of Kernighan and Richie s book 1988 con tains both standards and highlights the differences between them which to an experi enced programmer are minor and are almost entirely limited to type definitions The C language gains in brevity by leaving out all special purpose functions such as mathematical operators input output and text handling and instead making it easy to build and use program libraries containing working algorithms for these functions Thus for example where FORTRAN defines a SQRT operator to compute a square root C in stead assumes that the user will link the program with a mathematical library contain ing an implementation of an algorithm for computing the square root of a number C compilers as a result can be relatively small and many different compilers debuggers and programming aids are available for C even on microcomputers In recent years a preference has been shown for C an extension to the C language C adds to the language support for the concept of a class a necessary requirement for the use of object oriented programming methods Although object oriented program ming is possible in many other languages the proliferation of C compilers has made C the development language of choice especially because all C compilers include the ANSI C language as a subset with complete forward compatibility Students should note that effective C programm
21. s and even the number of arguments and their meanings were changed by individual software producers This led to a failure of CORE based pro grams to be very portable one of the principal advantages of using a standard in the first place The CORE system has not been considered for formal adoption as a standard by any ISO or ANSI committee and will not be in the future because most of its concepts are now embedded in GKS The purpose of using a graphics standard is to ensure that a program written using the standard will survive for new systems will be portable and will work independently of the specifics of implementation Using standards we sacrifice direct access to a device although GKS provides a high level of specialized access through the generalized device primitive and special escape sequences supported by the language binding The ability to support such a generic work environment is achieved by separating the graphics into the graphics system the language binding and device drivers Device drivers are computer programs that convert graphics operations into instruc tions that communicate directly with the hardware This requires a device driver for each particular device we are going to use so that users must add device drivers as new devices are acquired For the programmer the trade off is flexibility for drawing speed Device independence also means that the same map can appear and look the same on a printer on an electrostatic pl
22. tation applications particu larly in computer aided design CAD architecture and computer assisted manufactur ing CAM The Computer Graphics Interface CGI specifies a set of basic elements for the con trol of data exchange between the device independent and device dependent levels in a graphics system This involves a standardized virtual device interface VDI CGI can op erate as software to software in which case it transfers graphics from metafile to binding or vice versa As a software to hardware link CGI allows stored graphics to be output to any sup ported device independently of the graphics system which generated them Thus a map stored in CGI could be reproduced using the available display devices or retrieved for analysis via a computer programming link The most important function of CGI is to al low device specific graphics systems to produce output that is in accordance with stan dards and so to support a larger number of devices A large number of additional graphics systems exist at this higher level For window based applications the X windows system from M I T has gained widespread accep tance In addition standards for graphical rendering and even entirely open graphics based operating systems are now in the works It is important to remember that to be of use a standard should be both supported and used Programs written to any standard will Sec 15 4 References 303 not have to be rewritten as hardware
23. tead of the data structures that have been optimized for general purpose data as in database manage ment or numerical computing Some of the more interesting and challenging data struc tures that exist are those that are specifically designed to deal with map data Cartographic programs usually have a very different audience from general purpose pro grams Sec 15 3 Software Development Methods 299 5 3 SOFTWARE DEVELOPMENT METHODS 15 3 1 Graphics Standards and GKS Graphics are extremely important for analytical and computer cartography Using graph ics with a programming language requires some form of interaction between the program and the graphics This is accomplished using a language binding There are really two different types of graphics systems each with different ways of achieving a binding The first type is extremely device specific and involves a direct or indirect message to the graphics board the special purpose computer that controls the graphic display Many graphics programs use the memory addresses on the graphics card directly and send val ues to the locations that cause the display to react For example on an IBM PC using EGA graphics in mode 6 memory locations start ing at address B800 hexadecimal and occupying 640 by 200 binary pixels control the screen display A binary array written to these locations would automatically appear on the display in fact very quickly We are virtually talking directly to the ha
Download Pdf Manuals
Related Search