IIIInnnnddddiiiiaaaannnn HHHHiiiillllllll CCCC SSSSttttyyyylllleeee aaaannnndddd CCCCooooddddiiiinnnngggg SSSSttttaaaannnnddddaaaarrrrddddssss aaaassss aaaammmmeeeennnnddddeeeedddd ffffoooorrrr UUUU ooooffff TTTT ZZZZoooooooollllooooggggyyyy UUUUNNNNIIIIXXXX||||---- L.W. Cannon R.A. Elliott L.W. Kirchhoff J.H. Miller J.M. Milner R.W. Mitze E.P. Schan N.O. Whittington Bell Labs Henry Spencer Zoology Computer Systems University of Toronto _A_B_S_T_R_A_C_T This document is an annotated (by the last author) version of the original paper of the same title. It describes a set of coding standards and recommendations which are local standards for officially-supported UNIX programs. The scope is coding style, not functional organization. April 18, 1990 _________________________ |- UNIX is a trademark of Bell Laboratories. IIIInnnnddddiiiiaaaannnn HHHHiiiillllllll CCCC SSSSttttyyyylllleeee aaaannnndddd CCCCooooddddiiiinnnngggg SSSSttttaaaannnnddddaaaarrrrddddssss aaaassss aaaammmmeeeennnnddddeeeedddd ffffoooorrrr UUUU ooooffff TTTT ZZZZoooooooollllooooggggyyyy UUUUNNNNIIIIXXXX||||---- L.W. Cannon R.A. Elliott L.W. Kirchhoff J.H. Miller J.M. Milner R.W. Mitze E.P. Schan N.O. Whittington Bell Labs Henry Spencer Zoology Computer Systems University of Toronto _1. _I_n_t_r_o_d_u_c_t_i_o_n This document is a result of a committee formed at Indian Hill to establish a common set of coding standards and recommendations for the Indian Hill community. The scope of this work is the coding style, not the functional organization of programs. The standards in this document are not specific to ESS programming only1. We have tried to combine previous work [1,6] on C style into a uniform set of standards that should be appropriate for any project using C2. _________________________ |- UNIX is a trademark of Bell Laboratories. 1. In fact, they're pretty good general standards. ``To be clear is professional; not to be clear is unprofessional.'' - Sir Ernest Gowers. This document is presented unadulterated; U of T variations, comments, exceptions, etc. are presented in footnotes. 2. Of necessity, these standards cannot cover all situations. Experience and informed judgement count for much. Inexperienced programmers who encounter unusual situations should consult 1) code written by experienced C programmers following these rules, or 2) experienced C programmers. April 18, 1990 - 2 - _2. _F_i_l_e _O_r_g_a_n_i_z_a_t_i_o_n A file consists of various sections that should be separated by several blank lines. Although there is no max- imum length requirement for source files, files with more than about 1500 lines are cumbersome to deal with. The edi- tor may not have enough temp space to edit the file, compi- lations will go slower, etc. Since most of us use 300 baud terminals, entire rows of asterisks, for example, should be discouraged3. Also lines longer than 80 columns are not handled well by all terminals and should be avoided if pos- sible4. The suggested order of sections for a file is as fol- lows: 1. Any header file includes should be the first thing in the file. 2. Immediately after the includes5 should be a prologue that tells what is in that file. A description of the purpose of the objects in the files (whether they be functions, external data declarations or definitions, or something else) is more useful than a list of the object names. 3. Any typedefs and defines that apply to the file as a whole are next. 4. Next come the global (external) data declarations. If a set of defines applies to a particular piece of glo- bal data (such as a flags word), the defines should be immediately after the data declaration6. 5. The functions come last7. _________________________ 3. This is not a problem at U of T, or most other sensible places, but rows of asterisks are still annoying. 4. Excessively long lines which result from deep indenting are often a symptom of poorly-organized code. 5. A common variation, in both Bell code and ours, is to reverse the order of sections 1 and 2. This is an acceptable practice. 6. Such defines should be indented to put the _d_e_f_i_n_es one level deeper than the first keyword of the declaration to which they apply. 7. They should be in some sort of meaningful order. Top- down is generally better than bottom-up, and a ``breadth-first'' approach (functions on a similar April 18, 1990 - 3 - _2._1. _F_i_l_e _N_a_m_i_n_g _C_o_n_v_e_n_t_i_o_n_s UNIX requires certain suffix conventions for names of files to be processed by the _c_c command [5]8. The following suffixes are required: +o C source file names must end in ._c +o Assembler source file names must end in ._s In addition the following conventions are universally followed: +o Relocatable object file names end in ._o +o Include header file names end in ._h 9 or ._d +o Ldp10 specification file names end in ._b +o Yacc source file names end in ._y +o Lex source file names end in ._l _3. _H_e_a_d_e_r _F_i_l_e_s Header files are files that are included in other files prior to compilation by the C preprocessor. Some are defined at the system level like _s_t_d_i_o._h which must be included by any program using the standard I/O library. Header files are also used to contain data declarations and defines that are needed by more than one program11. Header _________________________ level of abstraction together) is preferred over depth-first (functions defined as soon as possible after their calls). Considerable judgement is called for here. If defining large numbers of essentially- independent utility functions, consider alphabetical order. 8. In addition to the suffix conventions given here, it is conventional to use `Makefile' (not `makefile') for the control file for _m_a_k_e and `README' for a summary of the contents of a directory or directory tree. 9. Preferred. An alternate convention that may be preferable in multi-language environments is to use the same suffix as an ordinary source file but with two periods instead of one (e.g. ``foo..c''). 10. No idea what this is. 11. Don't use absolute pathnames for header files. Use the <_n_a_m_e> construction for getting them from a standard place, or define them relative to the current April 18, 1990 - 4 - files should be functionally organized, i.e., declarations for separate subsystems should be in separate header files. Also, if a set of declarations is likely to change when code is ported from one machine to another, those declarations should be in a separate header file. Header files should not be nested. Some objects like typedefs and initialized data definitions cannot be seen twice by the compiler in one compilation. On non-UNIX sys- tems this is also true of uninitialized declarations without the _e_x_t_e_r_n keyword12. This can happen if include files are nested and will cause the compilation to fail. _4. _E_x_t_e_r_n_a_l _D_e_c_l_a_r_a_t_i_o_n_s External declarations should begin in column 1. Each declaration should be on a separate line. A comment describing the role of the object being declared should be included, with the exception that a list of defined con- stants do not need comments if the constant names are suffi- cient documentation. The comments should be tabbed so that they line up underneath each other13. Use the tab character (CTRL I if your terminal doesn't have a separate key) rather than blanks. For structure and union template declarations, each element should be alone on a line with a comment describing it. The opening brace ( { ) should be on the same line as the structure tag, and the closing brace should be alone on a line in column 1, i.e. struct boat { int wllength; /* water line length in feet */ int type; /* see below */ long sarea; /* sail area in square feet */ }; /* * defines for boat.type14 */ #define KETCH 1 #define YAWL 2 #define SLOOP 3 #define SQRIG 4 #define MOTOR 5 _________________________ directory. The ----IIII option of the C compiler is the best way to handle extensive private libraries of header files; it permits reorganizing the directory structure without having to alter source files. 12. It should be noted that declaring variables in a header file is often a poor idea. Frequently it is a symptom of poor partitioning of code between files. 13. So should the constant names and their defined values. April 18, 1990 - 5 - If an external variable is initialized15 the equal sign should not be omitted16. int x = 1; char *msg = "message"; struct boat winner = { 40, /* water line length */ YAWL, 600 /* sail area */ }; 17 _5. _C_o_m_m_e_n_t_s Comments that describe data structures, algorithms, etc., should be in block comment form with the opening /* in column one, a * in column 2 before each line of comment text18, and the closing */ in columns 2-3. _________________________ 14. These defines are better put right after the declaration of _t_y_p_e, within the _s_t_r_u_c_t declaration, with enough tabs after # to indent _d_e_f_i_n_e one level more than the structure member declarations. 15. Any variable whose initial value is important should be _e_x_p_l_i_c_i_t_l_y initialized, or at the very least should be commented to indicate that C's default initialization to 0 is being relied on. 16. The empty initializer, ``{}'', should never be used. Structure initializations should be fully parenthesized with braces. Constants used to initialize longs should be explicitly long. 17. In any file which is part of a larger whole rather than a self-contained program, maximum use should be made of the _s_t_a_t_i_c keyword to make functions and variables local to single files. Variables in particular should be accessible from other files only when there is a clear need that cannot be filled in another way. Such usages should be commented to make it clear that another file's variables are being used; the comment should name the other file. 18. Some automated program-analysis packages use a different character in this position as a marker for lines with specific items of information. In particular, a line with a `-' here in a comment preceding a function is sometimes assumed to be a one- line summary of the function's purpose. April 18, 1990 - 6 - /* * Here is a block comment. * The comment text should be tabbed over19 * and the opening /* and closing star-slash * should be alone on a line. */ Note that _g_r_e_p ^.\* will catch all block comments in the file. In some cases, block comments inside a function are appropriate, and they should be tabbed over to the same tab setting as the code that they describe. Short comments may appear on a single line indented over to the tab setting of the code that follows. if (argc > 1) { /* Get input file from command line. */ if (freopen(argv[1], "r", stdin) == NULL) error("can't open %s\n", argv[1]); } Very short comments may appear on the same line as the code they describe, but should be tabbed over far enough to separate them from the statements. If more than one short comment appears in a block of code they should all be tabbed to the same tab setting. if (a == 2) return(TRUE); /* special case */ else return(isprime(a)); /* works only for odd a */ _6. _F_u_n_c_t_i_o_n _D_e_c_l_a_r_a_t_i_o_n_s Each function should be preceded by a block comment prologue that gives the name and a short description of what the function does20. If the function returns a value, the type of the value returned should be alone on a line in column 1 (do not default to _i_n_t). If the function does not return a value then it should not be given a return type. _________________________ 19. A common practice in both Bell and local code is to use a space rather than a tab after the *. This is acceptable. 20. Discussion of non-trivial design decisions is also appropriate, but avoid duplicating information that is present in (and clear from) the code. It's too easy for such redundant information to get out of date. April 18, 1990 - 7 - If the value returned requires a long explanation, it should be given in the prologue; otherwise it can be on the same line as the return type, tabbed over. The function name and formal parameters should be alone on a line beginning in column 1. Each parameter should be declared (do not default to _i_n_t), with a comment on a single line. The opening brace of the function body should also be alone on a line begin- ning in column 1. The function name, argument declaration list, and opening brace should be separated by a blank line21. All local declarations and code within the function body should be tabbed over at least one tab. If the function uses any external variables, these should have their own declarations in the function body using the _e_x_t_e_r_n keyword. If the external variable is an array the array bounds must be repeated in the _e_x_t_e_r_n declaration. There should also be _e_x_t_e_r_n declarations for all functions called by a given function. This is particu- larly beneficial to someone picking up code written by another. If a function returns a value of type other than _i_n_t, it is required by the compiler that such functions be declared before they are used. Having the _e_x_t_e_r_n delcara- tion in the calling function's declarations section avoids all such problems22. In general each variable declaration should be on a separate line with a comment describing the role played by the variable in the function. If the variable is external or a parameter of type pointer which is changed by the func- tion, that should be noted in the comment. All such com- ments for parameters and local variables should be tabbed so that they line up underneath each other. The declarations should be separated from the function's statements by a blank line. A local variable should not be redeclared in nested blocks23. Even though this is valid C, the potential _________________________ 21. Neither Bell nor local code has ever included these separating blank lines, and it is not clear that they add anything useful. Leave them out. 22. These rules tend to produce a lot of clutter. Both Bell and local practice frequently omits _e_x_t_e_r_n declarations for _s_t_a_t_i_c variables and functions. This is permitted. Omission of declarations for standard library routines is also permissible, although if they _a_r_e declared it is better to declare them within the functions that use them rather than globally. 23. In fact, avoid any local declarations that override declarations at higher levels. April 18, 1990 - 8 - confusion is enough that _l_i_n_t will complain about it when given the ----hhhh option. _6._1. _E_x_a_m_p_l_e_s /* * skyblue() * * Determine if the sky is blue. */ int /* TRUE or FALSE */ skyblue() { extern int hour; if (hour < MORNING || hour > EVENING) return(FALSE); /* black */ else return(TRUE); /* blue */ } /* * tail(nodep) * * Find the last element in the linked list * pointed to by nodep and return a pointer to it. */ NODE * /* pointer to tail of list */ tail(nodep) NODE *nodep; /* pointer to head of list */ { register NODE *np; /* current pointer advances to NULL */ register NODE *lp; /* last pointer follows np */ np = lp = nodep; while ((np = np->next) != NULL) lp = np; return(lp); } _7. _C_o_m_p_o_u_n_d _S_t_a_t_e_m_e_n_t_s Compound statements are statements that contain lists of statements enclosed in braces. The enclosed list should be tabbed over one more than the tab position of the com- pound statement itself. The opening left brace should be at April 18, 1990 - 9 - the end of the line beginning the compound statement and the closing right brace should be alone on a line, tabbed under the beginning of the compound statement. Note that the left brace beginning a function body is the only occurrence of a left brace which is alone on a line. _7._1. _E_x_a_m_p_l_e_s if (expr) { statement; statement; } if (expr) { statement; statement; } else { statement; statement; } Note that the right brace before the _e_l_s_e and the right brace before the _w_h_i_l_e of a _d_o-_w_h_i_l_e statement (below) are the only places where a right braces appears that is not alone on a line. for (i = 0; i < MAX; i++) { statement; statement; } while (expr) { statement; statement; } do { statement; statement; } while (expr); switch (expr) { case ABC: case DEF: statement; break; case XYZ: statement; break; default: statement; break24; } April 18, 1990 - 10 - Note that when multiple _c_a_s_e labels are used, they are placed on separate lines. The fall through feature of the C _s_w_i_t_c_h statement should rarely if ever be used when code is executed before falling through to the next one. If this is done it must be commented for future maintenance. if (strcmp(reply, "yes") == EQUAL) { statements for yes ... } else if (strcmp(reply, "no") == EQUAL) { statements for no ... } else if (strcmp(reply, "maybe") == EQUAL) { statements for maybe ... } else { statements for none of the above ... } The last example is a generalized _s_w_i_t_c_h statement and the tabbing reflects the switch between exactly one of several alternatives rather than a nesting of statements. _8. _E_x_p_r_e_s_s_i_o_n_s _8._1. _O_p_e_r_a_t_o_r_s The old versions of equal-ops =+, =-, =*, etc. should not be used. The preferred use is +=, -=, *=, etc. All binary operators except . and -> should be separated from their operands by blanks25. In addition, keywords that are followed by expressions in parentheses should be separated from the left parenthesis by a blank26. Blanks should also appear after commas in argument lists to help separate the arguments visually. On the other hand, macros with argu- ments and function calls should not have a blank between the name and the left parenthesis. In particular, the C prepro- cessor requires the left parenthesis to be immediately after _________________________ 24. This _b_r_e_a_k is, strictly speaking, unnecessary, but it is required nonetheless because it prevents a fall- through error if another _c_a_s_e is added later after the last one. 25. Some judgement is called for in the case of complex expressions, which may be clearer if the ``inner'' operators are not surrounded by spaces and the ``outer'' ones are. 26. _S_i_z_e_o_f is an exception, see the discussion of function calls. Less logically, so is _r_e_t_u_r_n. April 18, 1990 - 11 - the macro name or else the argument list will not be recog- nized. Unary operators should not be separated from their single operand. Since C has some unexpected precedence rules, all expressions involving mixed operators should be fully parenthesized. _E_x_a_m_p_l_e_s a += c + d; a = (a + b) / (c * d); strp->field = str.fl - ((x & MASK) >> DISP); while (*d++ = *s++) ; /* EMPTY BODY */ _8._2. _N_a_m_i_n_g _C_o_n_v_e_n_t_i_o_n_s Individual projects will no doubt have their own naming conventions. There are some general rules however. +o An initial underscore should not be used for any user- created names27. UNIX uses it for names that the user should not have to know (like the standard I/O library)28. +o Macro names, _t_y_p_e_d_e_f names, and _d_e_f_i_n_e names should be all in CAPS. +o Variable names, structure tag names, and function names should be in lower case29. Some macros (such as _g_e_t_c_h_a_r and _p_u_t_c_h_a_r) are in lower case since they may also exist as functions. Care is needed when inter- changing macros and functions since functions pass their parameters by value whereas macros pass their arguments by name substitution30. _________________________ 27. Trailing underscores should be avoided too. 28. This convention is reserved for system purposes. If you must have your own private identifiers, begin them with a capital letter identifying the package to which they belong. 29. It is best to avoid names that differ only in case, like _f_o_o and _F_O_O. The potential for confusion is considerable. 30. This difference also means that carefree use of macros requires care when they are defined. Remember that complex expressions can be used as parameters, and operator-precedence problems can arise unless all occurrences of parameters in the definition have parentheses around them. There is little that can be April 18, 1990 - 12 - _8._3. _C_o_n_s_t_a_n_t_s Numerical constants should not be coded directly31. The _d_e_f_i_n_e feature of the C preprocessor should be used to assign a meaningful name. This will also make it easier to administer large programs since the constant value can be changed uniformly by changing only the _d_e_f_i_n_e. The enumera- tion data type is the preferred way to handle situations where a variable takes on only a discrete set of values, since additional type checking is available through _l_i_n_t. There are some cases where the constants 0 and 1 may appear as themselves instead of as defines. For example if a _f_o_r loop indexes through an array, then for (i = 0; i < ARYBOUND; i++) is reasonable while the code fptr = fopen(filename, "r"); if (fptr == 0) error("can't open %s\n", filename); is not. In the last example the defined constant _N_U_L_L is available as part of the standard I/O library's header file _s_t_d_i_o._h and must be used in place of the 0. _9. _P_o_r_t_a_b_i_l_i_t_y The advantages of portable code are well known. This section gives some guidelines for writing portable code, where the definition of portable is taken to mean that a source file contains portable code if it can be compiled and executed on different machines with the only source change being the inclusion of possibly different header files. The header files will contain defines and typedefs that may vary from machine to machine. Reference [1] contains useful information on both style and portability. Many of the recommendations in this document originated in [1]. The following is a list of pitfalls to be avoided and recommen- dations to be considered when designing portable code: +o First, one must recognize that some things are inherently non-portable. Examples are code to deal with particular hardware registers such as the program _________________________ done about the problems caused by side effects in parameters except to avoid side effects in expressions (a good idea anyway). 31. At the very least, any directly-coded numerical constant must have a comment explaining the derivation of the value. April 18, 1990 - 13 - status word, and code that is designed to support a particular piece of hardware such as an assembler or I/O driver. Even in these cases there are many rou- tines and data organizations that can be made machine independent. It is suggested that source file be organized so that the machine-independent code and the machine-dependent code are in separate files. Then if the program is to be moved to a new machine, it is a much easier task to determine what needs to be changed32. It is also possible that code in the machine-independent files may have uses in other pro- grams as well. +o Pay attention to word sizes. The following sizes apply to basic types in C for the machines that will be used most at IH33: type pdp11 3B IBM ________________________ char 8 8 8 short 16 16 16 int 16 32 32 long 32 32 32 In general if the word size is important, _s_h_o_r_t or _l_o_n_g should be used to get 16 or 32 bit items on any of the above machines34. If a simple loop counter is being used where either 16 or 32 bits will do, then use _i_n_t, since it will get the most efficient (natural) unit for the current machine35. _________________________ 32. If you #_i_f_d_e_f dependencies, make sure that if no machine is specified, the result is a syntax error, _n_o_t a default machine! 33. The 3B is a Bell Labs machine. The VAX, not shown in the table, is similar to the 3B in these respects. The 68000 resembles either the pdp11 or the 3B, depending on the particular compiler. 34. Any unsigned type other than plain _u_n_s_i_g_n_e_d _i_n_t should be _t_y_p_e_d_e_fed, as such types are highly compiler- dependent. This is also true of long and short types other than _l_o_n_g _i_n_t and _s_h_o_r_t _i_n_t. Large programs should have a central header file which supplies _t_y_p_e_d_e_fs for commonly-used width-sensitive types, to make it easier to change them and to aid in finding width-sensitive code. 35. Beware of making assumptions about the size of pointers. They are not always the same size as _i_n_t. Nor are all pointers always the same size, or freely interconvertible. Pointer-to-character is a particular April 18, 1990 - 14 - +o Word size also affects shifts and masks. The code x &= 0177770 will clear only the three rightmost bits of an _i_n_t on a PDP11. On a 3B it will also clear the entire upper halfword. Use x &= ~07 instead which works properly on all machines36. +o Code that takes advantage of the two's complement representation of numbers on most machines should not be used. Optimizations that replace arithmetic opera- tions with equivalent shifting operations are particu- larly suspect. You should weigh the time savings with the potential for obscure and difficult bugs when your code is moved, say, from a 3B to a 1A. +o Watch out for signed characters. On the PDP-11, char- acters are sign extended when used in expressions, which is not the case on any other machine. In partic- ular, _g_e_t_c_h_a_r is an integer-valued function (or macro) since the value of _E_O_F for the standard I/O library is -1, which is not possible for a character on the 3B or IBM37. +o The PDP-11 is unique among processors on which C exists in that the bytes are numbered from right to left within a word. All other machines (3B, IBM, Interdata 8/32, Honeywell) number the bytes from left to right38. Hence any code that depends on the left-right orienta- tion of bits in a word deserves special scrutiny. Bit fields within structure members will only be portable _________________________ trouble spot on machines which do not address to the byte. 36. The or operator ( | ) does not have these problems, nor do bitfields (which, unfortunately, are not very portable due to defective compilers). 37. Actually, this is not quite the real reason why _g_e_t_c_h_a_r returns _i_n_t, but the comment is valid: code which assumes either that characters are signed or that they are unsigned is unportable. It is best to completely avoid using _c_h_a_r to hold numbers. Manipulation of characters as if they were numbers is also often unportable. 38. Actually, there are some more right-to-left machines now, but the comments still apply. April 18, 1990 - 15 - so long as two separate fields are never concatenated and treated as a unit39. [1,3] +o Do not default the boolean test for non-zero, i.e. if (f() != FAIL) is better than if (f()) even though _F_A_I_L may have the value 0 which is con- sidered to mean false by C40. This will help you out later when somebody decides that a failure return should be -1 instead of 0 41. +o Be suspicious of numeric values appearing in the code. Even simple values like 0 or 1 could be better expressed using defines like _F_A_L_S_E and _T_R_U_E (see previ- ous item)42. Any other constants appearing in a pro- gram would be better expressed as a defined constant. This makes it easier to change and also easier to read. +o Become familiar with existing library functions and _________________________ 39. The same applies to variables in general. Alignment considerations and loader peculiarities make it very rash to assume that two consecutively-declared variables are together in memory, or that a variable of one type is aligned appropriately to be used as another type. 40. A particularly notorious case is using _s_t_r_c_m_p to test for string equality, where the result should _n_e_v_e_r _e_v_e_r be defaulted. The preferred approach is to define a macro _S_T_R_E_Q: #define STREQ(a, b) (strcmp((a), (b)) == 0) 41. An exception is commonly made for predicates, which are functions which meet the following restrictions: +o Has no other purpose than to return true or false. +o Returns 0 for false, 1 for true, nothing else. +o Is named so that the meaning of (say) a `true' return is absolutely obvious. Call a predicate _i_s_v_a_l_i_d or _v_a_l_i_d, not _c_h_e_c_k_v_a_l_i_d. 42. Actually, _Y_E_S and _N_O often read better. April 18, 1990 - 16 - defines43. You should not be writing your own string compare routine, or making your own defines for system structures44. Not only does this waste your time, but it prevents your program from taking advantage of any microcode assists or other means of improving perfor- mance of system routines45. +o Use _l_i_n_t. It is a valuable tool for finding machine- dependent constructs as well as other inconsistencies or program bugs that pass the compiler46. _1_0. _L_i_n_t _L_i_n_t is a C program checker [2] that examines C source files to detect and report type incompatibilities, incon- sistencies between function definitions and calls, potential program bugs, etc. It is expected that projects will require programs to use _l_i_n_t as part of the official accep- tance procedure47. In addition, work is going on in depart- ment 5521 to modify _l_i_n_t so that it will check for adherence to the standards in this document. It is still too early to say exactly which of the _________________________ 43. But not _t_o_o familiar. The internal details of library facilities, as opposed to their external interfaces, are subject to change without warning. They are also often quite unportable. 44. Or, especially, writing your own code to control terminals. Use the _t_e_r_m_c_a_p package. 45. It also makes your code less readable, because the reader has to figure out whether you're doing something special in that reimplemented stuff to justify its existence. Furthermore, it's a fruitful source of bugs. 46. The use of _l_i_n_t on all programs is strongly recommended. It is difficult to eliminate complaints about functions whose return value is not used (in the current version of C, at least), but most other messages from _l_i_n_t really do indicate something wrong. The -h, -p, -a, -x, and -c options are worth learning. All of them will complain about some legitimate things, but they will also pick up many botches. Note that -p checks function-call type-consistency for only a subset of Unix library routines, so programs should be linted both with and without this option for best ``coverage''. 47. Yes. April 18, 1990 - 17 - standards given here will be checked by _l_i_n_t. In some cases such as whether a comment is misleading or incorrect there is little hope of mechanical checking. In other cases such as checking that the opening brace of a function body is alone on a line in column 1, the test has already been added48. Future bulletins will be used to announce new additions to _l_i_n_t as they occur. It should be noted that the best way to use _l_i_n_t is not as a barrier that must be overcome before official accep- tance of a program, but rather as a tool to use whenever major changes or additions to the code have been made. _L_i_n_t can find obscure bugs and insure portability before problems occur. _1_1. _S_p_e_c_i_a_l _C_o_n_s_i_d_e_r_a_t_i_o_n_s This section contains some miscellaneous do's and don'ts. +o Don't change syntax via macro substitution. It makes the program unintelligible to all but the perpetrator. +o There is a time and a place for embedded assignment statements49. In some constructs there is no better way to accomplish the results without making the code bulkier and less readable. The _w_h_i_l_e loop in section 8.1 is one example of an appropriate place. Another is the common code segment: while ((c = getchar()) != EOF) { process the character } Using embedded assignment statements to improve run- time performance is also possible. However, one should consider the tradeoff between increased speed and decreased maintainability that results when embedded assignments are used in artificial places. For exam- ple, the code: a = b + c; d = a + r; should not be replaced by _________________________ 48. Little of this is relevant at U of T. The version of _l_i_n_t that we have lacks these mods. 49. The ++++++++ and -------- operators count as assignment statements. So, for many purposes, do functions with side effects. April 18, 1990 - 18 - d = (a = b + c) + r; even though the latter may save one cycle. Note that in the long run the time difference between the two will decrease as the optimizer gains maturity, while the difference in ease of maintenance will increase as the human memory of what's going on in the latter piece of code begins to fade50. +o There is also a time and place for the ternary ? : operator and the binary comma operator. The logical expression operand before the ? : should be parenthesized: (x >= 0) ? x : -x Nested ? : operators can be confusing and should be avoided if possible. There are some macros like _g_e_t_c_h_a_r where they can be useful. The comma operator can also be useful in _f_o_r statements to provide multi- ple initializations or incrementations. +o Goto statements should be used sparingly as in any well-structured code51. The main place where they can be usefully employed is to break out of several levels of _s_w_i_t_c_h, _f_o_r, and _w_h_i_l_e nesting52, e.g. for (...) for (...) { ... if (disaster) goto error; } ... error: clean up the mess When a _g_o_t_o is necessary the accompanying label should be alone on a line and tabbed one tab position to the _________________________ 50. Note also that side effects within expressions can result in code whose semantics are compiler-dependent, since C's order of evaluation is explicitly undefined in most places. Compilers do differ. 51. The _c_o_n_t_i_n_u_e statement is almost as bad. _B_r_e_a_k is less troublesome. 52. The need to do such a thing may indicate that the inner constructs should be broken out into a separate function, with a success/failure return code. April 18, 1990 - 19 - left of the associated code that follows. +o This committee recommends that programmers not rely on automatic beautifiers for the following reasons. First, the main person who benefits from good program style is the programmer himself. This is especially true in the early design of handwritten algorithms or pseudo-code. Automatic beautifiers can only be applied to complete, syntactically correct programs and hence are not available when the need for attention to white space and indentation is greatest. It is also felt that programmers can do a better job of making clear the complete visual layout of a function or file, with the normal attention to detail of a careful program- mer53. Sloppy programmers should learn to be careful programmers instead of relying on a beautifier to make their code readable. Finally, it is felt that since beautifiers are non-trivial programs that must parse the source, the burden of maintaining them in the face of the continuing evolution of C is not worth the bene- fits gained by such a program. _1_2. _P_r_o_j_e_c_t _D_e_p_e_n_d_e_n_t _S_t_a_n_d_a_r_d_s Individual projects may wish to establish additional standards beyond those given here. The following issues are some of those that should be adddressed by each project pro- gram administration group. +o What additional naming conventions should be followed? In particular, systematic prefix conventions for func- tional grouping of global data and also for structure or union member names can be useful. +o What kind of include file organization is appropriate for the project's particular data hierarchy? +o What procedures should be established for reviewing _l_i_n_t complaints? A tolerance level needs to be esta- blished in concert with the _l_i_n_t options to prevent unimportant complaints from hiding complaints about real bugs or inconsistencies. +o If a project establishes its own archive libraries, it should plan on supplying a lint library file [2] to the system administrators. This will allow _l_i_n_t to check for compatible use of library functions. _________________________ 53. In other words, some of the visual layout is dictated by intent rather than syntax. Beautifiers cannot read minds. April 18, 1990 - 20 - _1_3. _C_o_n_c_l_u_s_i_o_n A set of standards has been presented for C programming style. One of the most important points is the proper use of white space and comments so that the structure of the program is evident from the layout of the code. Another good idea to keep in mind when writing code is that it is likely that you or someone else will be asked to modify it or make it run on a different machine sometime in the future. As with any standard, it must be followed if it is to be useful. The Indian Hill version of _l_i_n_t will enforce those standards that are amenable to automatic checking. If you have trouble following any of these standards don't just ignore them. Programmers at Indian Hill should bring their problems to the Software Development System Group (Lee Kirchhoff, contact) in department 5522. Programmers outside Indian Hill should contact the Processor Application Group (Layne Cannon, contact) in department 5512 54. _________________________ 54. At U of T Zoology, it's Henry Spencer in 336B. April 18, 1990 - 21 - RRRReeeeffffeeeerrrreeeennnncccceeeessss [1] B.A. Tague, "C Language Portability", Sept 22, 1977. This document issued by department 8234 contains three memos by R.C. Haight, A.L. Glasser, and T.L. Lyon deal- ing with style and portability. [2] S.C. Johnson, "Lint, a C Program Checker", Technical Memorandum, 77-1273-14, September 16, 1977. [3] R.W. Mitze, "The 3B/PDP-11 Swabbing Problem", Memoran- dum for File, 1273-770907.01MF, September 14, 1977. [4] R.A. Elliott and D.C. Pfeffer, "3B Processor Common Diagnostic Standards- Version 1", Memorandum for File, 5514-780330.01MF, March 30, 1978. [5] R.W. Mitze, "An Overview of C Compilation of UNIX User Processes on the 3B", Memorandum for File, 5521- 780329.02MF, March 29, 1978. [6] B.W. Kernighan and D.M. Ritchie, _T_h_e _C _P_r_o_g_r_a_m_m_i_n_g _L_a_n_g_u_a_g_e, Prentice-Hall 1978. April 18, 1990 - 22 - /* * TTTThhhheeee CCCC SSSSttttyyyylllleeee SSSSuuuummmmmmmmaaaarrrryyyy SSSShhhheeeeeeeetttt Block comment, * by Henry Spencer, U of T Zoology describes file. */ #include Headers; don't nest. typedef int SEQNO; /* ... */ Global definitions. #define STREQ(a, b) (strcmp((a), (b)) == 0) static char *foo = NULL; /* ... */ Global declarations. struct bar { Static whenever poss. SEQNO alpha; /* ... */ # define NOSEQNO 0 int beta; /* ... */ Don't assume 16 bits. }; /* * Many unnecessary braces, to show where. Functions. */ static int /* what is returned */ Don't default int. bletch(a) int a; /* ... */ Don't default int. { int bar; /* ... */ extern int errno; /* ..., changed here */ extern char *index(); if (foobar() != FAIL) { if (!isvalid()) { return(OK); errno = ERANGE; } } else { x = &y + z->field; while (x == (y & MASK)) { } f += (x >= 0) ? x : -x; } for (i = 0; i < BOUND; i++) { /* lint -h[p]cax. */ do { } /* Avoid nesting ?: */ } while (index(a, b) != NULL); if (STREQ(x, "foo")) { x |= 07; /* 07 is... */ switch (...) { } else if (STREQ(x, "bar")) { case ABC: x &= ~077; /* 077 is... */ case DEF: } else if (STREQ(x, "ugh")) { printf("...", a, b); /* Avoid gotos */ break; } else { case XYZ: /* and continues. */ x = y; } /* FALLTHROUGH */ default: while ((c = getc()) != EOF) /* Limit imbedded =s. */ ; /* NULLBODY */ break; } } April 18, 1990 - 23 - --------------- April 18, 1990