2 <HEAD><TITLE>APR Design Document</TITLE></HEAD>
6 <p>The Apache Portable Run-time libraries have been designed to provide a common
7 interface to low level routines across any platform. The original goal of APR
8 was to combine all code in Apache to one common code base. This is not the
9 correct approach however, so the goal of APR has changed. There are places
10 where common code is not a good thing. For example, how to map requests
11 to either threads or processes should be platform specific. APR's place
12 is now to combine any code that can be safely combined without sacrificing
15 <p>To this end we have created a set of operations that are required for cross
16 platform development. There may be other types that are desired and those
17 will be implemented in the future.</p>
19 <p>This document will discuss the structure of APR, and how best to contribute
20 code to the effort.</p>
22 <h2>APR On Windows and Netware</h2>
24 <p>APR on Windows and Netware is different from APR on all other systems,
25 because those platforms don't use autoconf. On Unix, apr_private.h (private to
26 APR) and apr.h (public, used by applications that use APR) are generated by
27 autoconf from acconfig.h and apr.h.in respectively. On Windows (and Netware),
28 apr_private.h and apr.h are created from apr_private.hw (apr_private.hwn)
29 and apr.hw (apr.hwn) respectively.</p>
32 If you add code to acconfig.h or tests to configure.in or aclocal.m4,
33 please give some thought to whether or not Windows and Netware need
34 these additions as well. A general rule of thumb, is that if it is
35 a feature macro, such as APR_HAS_THREADS, Windows and Netware need it.
36 In other words, if the definition is going to be used in a public APR
37 header file, such as apr_general.h, Windows needs it.
39 The only time it is safe to add a macro or test without also adding
40 the macro to apr*.h[n]w, is if the macro tells APR how to build. For
41 example, a test for a header file does not need to be added to Windows.
46 <p>One of the goals of APR is to provide a common set of features across all
47 platforms. This is an admirable goal, it is also not realistic. We cannot
48 expect to be able to implement ALL features on ALL platforms. So we are
49 going to do the next best thing. Provide a common interface to ALL APR
50 features on MOST platforms.</p>
52 <p>APR developers should create FEATURE MACROS for any feature that is not
53 available on ALL platforms. This should be a simple definition which has
56 <code>APR_HAS_FEATURE</code>
58 <p>This macro should evaluate to true if APR has this feature on this platform.
59 For example, Linux and Windows have mmap'ed files, and APR is providing an
60 interface for mmapp'ing a file. On both Linux and Windows, APR_HAS_MMAP
61 should evaluate to one, and the ap_mmap_* functions should map files into
62 memory and return the appropriate status codes.</p>
64 <p>If your OS of choice does not have mmap'ed files, APR_HAS_MMAP should
65 evaluate to zero, and all ap_mmap_* functions should not be defined. The
66 second step is a precaution that will allow us to break at compile time if a
67 programmer tries to use unsupported functions.</p>
71 <p>The base types in APR</p>
75 Shared library routines
85 Process and thread locks (critical sections)
89 File I/O, including pipes
91 Atomic integer operations
93 String handling routines
95 Pool-based memory allocation
97 Reading passwords from the terminal
103 Threads and processes
105 Any APR type which doesn't have any other place to belong. This
106 should be used sparingly.
108 Functions meant to be used across multiple APR types. This area
109 is for internal functions only. If a function is exposed, it should
113 <h2>Directory Structure</h2>
115 <p>Each type has a base directory. Inside this base directory, are
116 subdirectories, which contain the actual code. These subdirectories are named
117 after the platforms the are compiled on. Unix is also used as a common
118 directory. If the code you are writing is POSIX based, you should look at the
119 code in the unix directory. A good rule of thumb, is that if more than half
120 your code needs to be ifdef'ed out, and the structures required for your code
121 are substantively different from the POSIX code, you should create a new
124 <p>Currently, the APR code is written for Unix, BeOS, Windows, and OS/2. An
125 example of the directory structure is the file I/O directory:</p>
132 -> unix The Unix and common base code
134 -> win32 The Windows code
139 <p>Obviously, BeOS does not have a directory. This is because BeOS is currently
140 using the Unix directory for it's file_io.</p>
142 <p>There are a few special top level directories. These are test and include.
143 Test is a directory which stores all test programs. It is expected
144 that if a new type is developed, there will also be a new test program, to
145 help people port this new type to different platforms. A small document
146 describing how to create new tests that integrate with the test suite can be
147 found in the test/ directory. Include is a directory which stores all
148 required APR header files for external use.</p>
150 <h2>Creating an APR Type</h2>
152 <p>The current design of APR requires that most APR types be incomplete.
153 It is not possible to write flexible portable code if programs can access
154 the internals of APR types. This is because different platforms are
155 likely to define different native types. There are only two execptions to
159 <li>The first exception to this rule is if the type can only reasonably be
160 implemented one way. For example, time is a complete type because there
161 is only one reasonable time implementation.
163 <li>The second exception to the incomplete type rule can be found in
164 apr_portable.h. This file defines the native types for each platform.
165 Using these types, it is possible to extract native types for any APR type.</p>
168 <p>For this reason, each platform defines a structure in their own directories.
169 Those structures are then typedef'ed in an external header file. For example
170 in file_io/unix/fileio.h:</p>
181 <p>In include/apr_file_io.h:</p>
183 typedef struct ap_file_t ap_file_t;
186 <p> This will cause a compiler error if somebody tries to access the filedes
187 field in this structure. Windows does not have a filedes field, so obviously,
188 it is important that programs not be able to access these.</p>
190 <p>You may notice the apr_pool_t field. Most APR types have this field. This
191 type is used to allocate memory within APR. Because every APR type has a pool,
192 any APR function can allocate memory if it needs to. This is very important
193 and it is one of the reasons that APR works. If you create a new type, you
194 must add a pool to it. If you do not, then all functions that operate on that
195 type will need a pool argument.</p>
197 <h2>New Function</h2>
199 <p>When creating a new function, please try to adhere to these rules.</p>
202 <li> Result arguments should be the first arguments.
203 <li> If a function needs a pool, it should be the last argument.
204 <li> These rules are flexible, especially if it makes the code easier
205 to understand because it mimics a standard function.
208 <h2>Documentation</h2>
210 <p>Whenever a new function is added to APR, it MUST be documented. New
211 functions will not be committed unless there are docs to go along with them.
212 The documentation should be a comment block above the function in the header
215 <p>The format for the comment block is:</p>
219 * Brief description of the function
220 * @param parma_1_name explanation
221 * @param parma_2_name explanation
222 * @param parma_n_name explanation
223 * @tip Any extra information people should know.
224 * @deffunc function prototype if required
228 <p>For an actual example, look at any file in the include directory. The
229 reason the docs are in the header files is to ensure that the docs always
230 reflect the current code. If you change paramters or return values for a
231 function, please be sure to update the documentation.</p>
233 <h2>APR Error reporting</h2>
235 <p>Most APR functions should return an ap_status_t type. The only time an
236 APR function does not return an ap_status_t is if it absolutely CAN NOT
237 fail. Examples of this would be filling out an array when you know you are
238 not beyond the array's range. If it cannot fail on your platform, but it
239 could conceivably fail on another platform, it should return an ap_status_t.
240 Unless you are sure, return an ap_status_t.</p>
243 This includes functions that return TRUE/FALSE values. How that
244 is handled is discussed below
247 <p>All platforms return errno values unchanged. Each platform can also have
248 one system error type, which can be returned after an offset is added.
249 There are five types of error values in APR, each with it's own offset.</p>
251 <!-- This should be turned into a table, but I am lazy today -->
254 0) This is 0 for all platforms and isn't really defined
255 anywhere, but it is the offset for errno values.
256 (This has no name because it isn't actually defined,
257 but for completeness we are discussing it here).
259 1) APR_OS_START_ERROR This is platform dependent, and is the offset at which
260 APR errors start to be defined. Error values are
261 defined as anything which caused the APR function to
262 fail. APR errors in this range should be named
263 APR_E* (i.e. APR_ENOSOCKET)
265 2) APR_OS_START_STATUS This is platform dependent, and is the offset at which
266 APR status values start. Status values do not indicate
267 success or failure, and should be returned if
268 APR_SUCCESS does not make sense. APR status codes in
269 this range should be name APR_* (i.e. APR_DETACH)
271 4) APR_OS_START_USEERR This is platform dependent, and is the offset at which
272 APR apps can begin to add their own error codes.
274 3) APR_OS_START_SYSERR This is platform dependent, and is the offset at which
275 system error values begin.
278 <strong>The difference in naming between APR_OS_START_ERROR and
279 APR_OS_START_STATUS mentioned above allows programmers to easily determine if
280 the error code indicates an error condition or a status codition.</strong>
282 <p>If your function has multiple return codes that all indicate success, but
283 with different results, or if your function can only return PASS/FAIL, you
284 should still return an apr_status_t. In the first case, define one
285 APR status code for each return value, an example of this is
286 <code>apr_proc_wait</code>, which can only return APR_CHILDDONE,
287 APR_CHILDNOTDONE, or an error code. In the second case, please return
288 APR_SUCCESS for PASS, and define a new APR status code for failure, an
289 example of this is <code>apr_compare_users</code>, which can only return
290 APR_SUCCESS, APR_EMISMATCH, or an error code.</p>
292 <p>All of these definitions can be found in apr_errno.h for all platforms. When
293 an error occurs in an APR function, the function must return an error code.
294 If the error occurred in a system call and that system call uses errno to
295 report an error, then the code is returned unchanged. For example: </p>
298 if (open(fname, oflags, 0777) < 0)
302 <p>The next place an error can occur is a system call that uses some error value
303 other than the primary error value on a platform. This can also be handled
304 by APR applications. For example:</p>
307 if (CreateFile(fname, oflags, sharemod, NULL,
308 createflags, attributes, 0) == INVALID_HANDLE_VALUE
309 return (GetLAstError() + APR_OS_START_SYSERR);
312 <p>These two examples implement the same function for two different platforms.
313 Obviously even if the underlying problem is the same on both platforms, this
314 will result in two different error codes being returned. This is OKAY, and
315 is correct for APR. APR relies on the fact that most of the time an error
316 occurs, the program logs the error and continues, it does not try to
317 programatically solve the problem. This does not mean we have not provided
318 support for programmatically solving the problem, it just isn't the default
319 case. We'll get to how this problem is solved in a little while.</p>
321 <p>If the error occurs in an APR function but it is not due to a system call,
322 but it is actually an APR error or just a status code from APR, then the
323 appropriate code should be returned. These codes are defined in apr_errno.h
324 and should be self explanatory.</p>
326 <p>No APR code should ever return a code between APR_OS_START_USEERR and
327 APR_OS_START_SYSERR, those codes are reserved for APR applications.</p>
329 <p>To programmatically correct an error in a running application, the error
330 codes need to be consistent across platforms. This should make sense. APR
331 has provided macros to test for status code equivalency. For example, to
332 determine if the code that you received from the APR function means EOF, you
333 would use the macro APR_STATUS_IS_EOF().</p>
335 <p>Why did APR take this approach? There are two ways to deal with error
339 <li> Return the same error code across all platforms.
340 <li> Return platform specific error codes and convert them when necessary.
343 <p>The problem with option number one is that it takes time to convert error
344 codes to a common code, and most of the time programs want to just output
345 an error string. If we convert all errors to a common subset, we have four
346 steps to output an error string:</p>
348 <p>The seocnd problem with option 1, is that it is a lossy conversion. For
349 example, Windows and OS/2 have a couple hundred error codes, but POSIX errno
350 only defines about 50 errno values. This means that if we convert to a
351 canonical error value immediately, there is no way for the programmer to
352 get the actual system error.</p>
355 make syscall that fails
356 convert to common error code step 1
357 return common error code
359 call error output function step 2
360 convert back to system error step 3
361 output error string step 4
364 <p>By keeping the errors platform specific, we can output error strings in two
368 make syscall that fails
371 call error output function step 1
372 output error string step 2
375 <p>Less often, programs change their execution based on what error was returned.
376 This is no more expensive using option 2 than it is using option 1, but we
377 put the onus of converting the error code on the programmer themselves.
378 For example, using option 1:</p>
381 make syscall that fails
382 convert to common error code
383 return common error code
384 decide execution based on common error code
387 <p>Using option 2:</p>
390 make syscall that fails
392 convert to common error code (using ap_canonical_error)
393 decide execution based on common error code
396 <p>Finally, there is one more operation on error codes. You can get a string
397 that explains in human readable form what has happened. To do this using
398 APR, call ap_strerror().</p>