John R. Sheets | 1e8e5ba | 2000-08-08 01:24:00 +0000 | [diff] [blame] | 1 | <chapter id="debugging"> |
John R. Sheets | d9e064f | 2000-12-13 21:52:37 +0000 | [diff] [blame] | 2 | <title>Debug Logging</title> |
John R. Sheets | 1e8e5ba | 2000-08-08 01:24:00 +0000 | [diff] [blame] | 3 | |
| 4 | <para> |
John R. Sheets | d9e064f | 2000-12-13 21:52:37 +0000 | [diff] [blame] | 5 | Written by &name-dimitrie-paun; <email>&email-dimitrie-paun;</email>, 28 Mar 1998 |
John R. Sheets | 1e8e5ba | 2000-08-08 01:24:00 +0000 | [diff] [blame] | 6 | </para> |
| 7 | <para> |
| 8 | (Extracted from <filename>wine/documentation/debug-msgs</filename>) |
| 9 | </para> |
| 10 | |
| 11 | <note> |
| 12 | <para> |
| 13 | The new debugging interface can be considered to be |
| 14 | stable, with the exception of the in-memory message |
| 15 | construction functions. However, there is still a lot of |
| 16 | work to be done to polish things up. To make my life |
| 17 | easier, please follow the guidelines described in this |
| 18 | document. |
| 19 | </para> |
| 20 | </note> |
| 21 | |
| 22 | <important> |
| 23 | <para> |
| 24 | Read this document before writing new code. DO NOT USE |
| 25 | <function>fprintf</function> (or |
| 26 | <function>printf</function>) to output things. Also, instead |
| 27 | of writing FIXMEs in the source, output a FIXME message if |
| 28 | you can. |
| 29 | </para> |
| 30 | <para> |
| 31 | At the end of the document, there is a "Style Guide" for |
| 32 | debugging messages. Please read it. |
| 33 | </para> |
| 34 | </important> |
| 35 | |
John R. Sheets | d9e064f | 2000-12-13 21:52:37 +0000 | [diff] [blame] | 36 | <sect1 id="dbg-classes"> |
John R. Sheets | 1e8e5ba | 2000-08-08 01:24:00 +0000 | [diff] [blame] | 37 | <title>Debugging classes</title> |
| 38 | |
| 39 | <para> |
| 40 | There are 4 types (or classes) of debugging messages: |
| 41 | </para> |
| 42 | <variablelist> |
| 43 | <varlistentry> |
| 44 | <term><literal>FIXME</literal></term> |
| 45 | <listitem> |
| 46 | <para> |
| 47 | Messages in this class relate to behavior of Wine that |
| 48 | does not correspond to standard Windows behavior and |
| 49 | that should be fixed. |
| 50 | </para> |
| 51 | <para>Examples: stubs, semi-implemented features, etc.</para> |
| 52 | </listitem> |
| 53 | </varlistentry> |
| 54 | <varlistentry> |
| 55 | <term><literal>ERR</literal></term> |
| 56 | <listitem> |
| 57 | <para> |
| 58 | Messages in this class relate to serious errors in |
| 59 | Wine. This sort of messages are close to asserts -- |
| 60 | that is, you should output an error message when the |
| 61 | code detects a condition which should not happen. In |
| 62 | other words, important things that are not warnings |
| 63 | (see below), are errors. |
| 64 | </para> |
| 65 | <para> |
| 66 | Examples: unexpected change in internal state, etc. |
| 67 | </para> |
| 68 | </listitem> |
| 69 | </varlistentry> |
| 70 | <varlistentry> |
| 71 | <term><literal>WARN</literal></term> |
| 72 | <listitem> |
| 73 | <para> |
| 74 | These are warning messages. You should report a |
| 75 | warning when something unwanted happen but the |
| 76 | function behaves properly. That is, output a warning |
| 77 | when you encounter something unexpected (ex: could not |
| 78 | open a file) but the function deals correctly with the |
| 79 | situation (that is, according to the docs). If you do |
| 80 | not deal correctly with it, output a fixme. |
| 81 | </para> |
| 82 | <para> |
| 83 | Examples: fail to access a resource required by the |
| 84 | app, etc. |
| 85 | </para> |
| 86 | </listitem> |
| 87 | </varlistentry> |
| 88 | <varlistentry> |
| 89 | <term><literal>TRACE</literal></term> |
| 90 | <listitem> |
| 91 | <para> |
| 92 | These are detailed debugging messages that are mainly |
| 93 | useful to debug a component. These are usually turned |
| 94 | off. |
| 95 | </para> |
| 96 | <para> |
| 97 | Examples: everything else that does not fall in one of |
| 98 | the above mentioned categories and the user does not |
| 99 | need to know about it. |
| 100 | </para> |
| 101 | </listitem> |
| 102 | </varlistentry> |
| 103 | </variablelist> |
| 104 | |
| 105 | <para> |
| 106 | The user has the capability to turn on or off messages of a |
| 107 | particular type. You can expect the following patterns of |
| 108 | usage (but note that any combination is possible): |
| 109 | </para> |
| 110 | <itemizedlist> |
| 111 | <listitem> |
| 112 | <para> |
| 113 | when you debug a component, all types |
| 114 | (<literal>TRACE</literal>, <literal>WARN</literal>, |
| 115 | <literal>ERR</literal>, <literal>FIXME</literal>) will |
| 116 | be enabled. |
| 117 | </para> |
| 118 | </listitem> |
| 119 | <listitem> |
| 120 | <para> |
| 121 | during the pre-alpha (maybe alpha) stage of Wine, most |
| 122 | likely the <literal>TRACE</literal> class will be |
| 123 | disabled by default, but all others |
| 124 | (<literal>WARN</literal>, <literal>ERR</literal>, |
| 125 | <literal>FIXME</literal>) will be enabled by default. |
| 126 | </para> |
| 127 | </listitem> |
| 128 | <listitem> |
| 129 | <para> |
| 130 | when Wine will become stable, most likely the |
| 131 | <literal>TRACE</literal> and <literal>WARN</literal> |
| 132 | classes will be disabled by default, but all |
| 133 | <literal>ERR</literal>s and <literal>FIXME</literal>s |
| 134 | will be enabled. |
| 135 | </para> |
| 136 | </listitem> |
| 137 | <listitem> |
| 138 | <para> |
| 139 | in some installations that want the smallest footprint |
| 140 | and where the debug information is of no interest, all |
| 141 | classes may be disabled by default. |
| 142 | </para> |
| 143 | </listitem> |
| 144 | </itemizedlist> |
| 145 | <para> |
| 146 | Of course, the user will have the runtime ability to |
| 147 | override these defaults. However, this ability may be turned |
| 148 | off and certain classes of messages may be completely |
| 149 | disabled at compile time to reduce the size of Wine. |
| 150 | </para> |
John R. Sheets | d9e064f | 2000-12-13 21:52:37 +0000 | [diff] [blame] | 151 | </sect1> |
John R. Sheets | 1e8e5ba | 2000-08-08 01:24:00 +0000 | [diff] [blame] | 152 | |
John R. Sheets | d9e064f | 2000-12-13 21:52:37 +0000 | [diff] [blame] | 153 | <sect1 id="dbg-channels"> |
John R. Sheets | 1e8e5ba | 2000-08-08 01:24:00 +0000 | [diff] [blame] | 154 | <title>Debugging channels</title> |
| 155 | |
| 156 | <para> |
| 157 | Also, we divide the debugging messages on a component basis. |
| 158 | Each component is assigned a debugging channel. The |
| 159 | identifier of the channel must be a valid C identifier but |
| 160 | note that it may also be a reserved word like |
| 161 | <type>int</type> or <type>static</type>. |
| 162 | </para> |
| 163 | <para> |
| 164 | Examples of debugging channels: |
| 165 | <simplelist type="inline"> |
| 166 | <member><literal>reg</literal></member> |
| 167 | <member><literal>updown</literal></member> |
| 168 | <member><literal>string</literal></member> |
| 169 | </simplelist> |
| 170 | </para> |
| 171 | <para> |
| 172 | We will refer to a generic channel as <literal>xxx</literal>. |
| 173 | </para> |
| 174 | <note> |
| 175 | <para> |
| 176 | for those who know the old interface, the channel/type is |
| 177 | what followed the _ in the |
| 178 | <function>dprintf_xxx</function> statements. For example, |
| 179 | to output a message on the debugging channel |
| 180 | <literal>reg</literal> in the old interface you would had |
| 181 | to write: |
| 182 | </para> |
| 183 | <programlisting> |
| 184 | dprintf_reg(stddeb, "Could not access key!\n"); |
| 185 | </programlisting> |
| 186 | <para> |
| 187 | In the new interface, we drop the |
| 188 | <literal>stddeb</literal> as it is implicit. However, we |
| 189 | add an orthogonal piece of information to the message: its |
| 190 | class. This is very important as it will allow us to |
| 191 | selectively turn on or off certain messages based on the |
| 192 | type of information they report. For this reason it is |
| 193 | essential to choose the right class for the message. |
| 194 | Anyhow, suppose we figured that this message should belong |
| 195 | in the <literal>WARN</literal> class, so in the new |
| 196 | interface, you write: |
| 197 | </para> |
| 198 | <programlisting> |
| 199 | WARN(reg, "Could not access key!\n"); |
| 200 | </programlisting> |
| 201 | </note> |
John R. Sheets | d9e064f | 2000-12-13 21:52:37 +0000 | [diff] [blame] | 202 | </sect1> |
John R. Sheets | 1e8e5ba | 2000-08-08 01:24:00 +0000 | [diff] [blame] | 203 | |
John R. Sheets | d9e064f | 2000-12-13 21:52:37 +0000 | [diff] [blame] | 204 | <sect1 id="dbg-using"> |
John R. Sheets | 1e8e5ba | 2000-08-08 01:24:00 +0000 | [diff] [blame] | 205 | <title>How to use it</title> |
| 206 | |
| 207 | <para> |
| 208 | So, to output a message (class <literal>YYY</literal>) on |
| 209 | channel <literal>xxx</literal>, do: |
| 210 | </para> |
| 211 | <programlisting> |
| 212 | #include "debug.h" |
| 213 | |
| 214 | .... |
| 215 | |
John R. Sheets | d9e064f | 2000-12-13 21:52:37 +0000 | [diff] [blame] | 216 | YYY(xxx, "<message>", ...); |
John R. Sheets | 1e8e5ba | 2000-08-08 01:24:00 +0000 | [diff] [blame] | 217 | </programlisting> |
| 218 | <para> |
| 219 | Some examples from the code: |
| 220 | </para> |
| 221 | <programlisting> |
| 222 | #include "debug.h" |
| 223 | |
| 224 | ... |
| 225 | |
| 226 | TRACE(crtdll, "CRTDLL_setbuf(file %p buf %p)", file, buf); |
| 227 | |
| 228 | WARN(aspi, "Error opening device errno=%d", save_error); |
| 229 | </programlisting> |
| 230 | <para> |
| 231 | If you need to declare a new debugging channel, use it in |
| 232 | your code and then do: |
| 233 | </para> |
| 234 | <screen> |
| 235 | %tools/make_debug |
| 236 | </screen> |
| 237 | <para> |
| 238 | in the root directory of Wine. Note that this will result in |
| 239 | almost complete recompilation of Wine. |
| 240 | </para> |
| 241 | |
| 242 | <note> |
| 243 | <orderedlist> |
| 244 | <listitem> |
| 245 | <para> |
| 246 | Please pay attention to which class you assign the |
| 247 | message. There are only 4 classes, so it is not hard. |
| 248 | The reason it is important to get it right is that too |
| 249 | much information is no information. For example, if |
| 250 | you put things into the <literal>WARN</literal> class |
| 251 | that should really be in the <literal>TRACE</literal> |
| 252 | class, the output will be too big and this will force |
| 253 | the user to turn warnings off. But this way he will |
| 254 | fail to see the important ones. Also, if you put |
| 255 | warnings into the <literal>TRACE</literal> class lets |
| 256 | say, he will most likely miss those because usually |
| 257 | the <literal>TRACE</literal> class is turned off. A |
| 258 | similar argument can be made if you mix any other two |
| 259 | classes. |
| 260 | </para> |
| 261 | </listitem> |
| 262 | <listitem> |
| 263 | <para> |
| 264 | All lines should end with a newline. If you can NOT |
| 265 | output everything that you want in the line with only |
| 266 | one statement, then you need to build the string in |
| 267 | memory. Please read the section below "In-memory |
| 268 | messages" on the preferred way to do it. PLEASE USE |
| 269 | THAT INTERFACE TO BUILD MESSAGES IN MEMORY. The reason |
| 270 | is that we are not sure that we like it and having |
| 271 | everything in one format will facilitate the |
| 272 | (automatic) translation to a better interface. |
| 273 | </para> |
| 274 | </listitem> |
| 275 | </orderedlist> |
| 276 | </note> |
John R. Sheets | d9e064f | 2000-12-13 21:52:37 +0000 | [diff] [blame] | 277 | </sect1> |
John R. Sheets | 1e8e5ba | 2000-08-08 01:24:00 +0000 | [diff] [blame] | 278 | |
John R. Sheets | d9e064f | 2000-12-13 21:52:37 +0000 | [diff] [blame] | 279 | <sect1 id="dbg-checking"> |
John R. Sheets | 1e8e5ba | 2000-08-08 01:24:00 +0000 | [diff] [blame] | 280 | <title>Are we debugging?</title> |
| 281 | |
| 282 | <para> |
| 283 | To test whether the debugging output of class |
| 284 | <literal>yyy</literal> on channel <literal>xxx</literal> is |
| 285 | enabled, use: |
| 286 | </para> |
| 287 | <screen> |
| 288 | TRACE_ON to test if TRACE is enabled |
| 289 | WARN_ON to test if WARN is enabled |
| 290 | FIXME_ON to test if FIXME is enabled |
| 291 | ERR_ON to test if ERR is enabled |
| 292 | </screen> |
| 293 | <para> |
| 294 | Examples: |
| 295 | </para> |
| 296 | <programlisting> |
| 297 | if(TRACE_ON(atom)){ |
| 298 | ...blah... |
| 299 | } |
| 300 | </programlisting> |
| 301 | |
| 302 | <note> |
| 303 | <para> |
| 304 | You should normally need to test only if |
| 305 | <literal>TRACE_ON</literal>. At present, none of the other |
| 306 | 3 tests (except for <literal>ERR_ON</literal> which is |
| 307 | used only once!) are used in Wine. |
| 308 | </para> |
| 309 | </note> |
John R. Sheets | d9e064f | 2000-12-13 21:52:37 +0000 | [diff] [blame] | 310 | </sect1> |
John R. Sheets | 1e8e5ba | 2000-08-08 01:24:00 +0000 | [diff] [blame] | 311 | |
John R. Sheets | d9e064f | 2000-12-13 21:52:37 +0000 | [diff] [blame] | 312 | <sect1 id="dbg-in-memory"> |
John R. Sheets | 1e8e5ba | 2000-08-08 01:24:00 +0000 | [diff] [blame] | 313 | <title>In-memory messages</title> |
| 314 | |
| 315 | <para> |
| 316 | If you NEED to build the message from multiple calls, you |
| 317 | need to build it in memory. To do that, you should use the |
| 318 | following interface: |
| 319 | </para> |
| 320 | |
| 321 | <orderedlist> |
| 322 | <listitem> |
| 323 | <para> |
| 324 | declare a string (where you are allowed to declare C |
| 325 | variables) as follows: |
| 326 | <programlisting> |
| 327 | dbg_decl_str(name, len); |
| 328 | </programlisting> |
| 329 | where <parameter>name</parameter> is the name of the |
| 330 | string (you should use the channel name on which you |
| 331 | are going to output it) |
| 332 | </para> |
| 333 | </listitem> |
| 334 | <listitem> |
| 335 | <para> |
| 336 | print in it with: |
| 337 | <programlisting> |
John R. Sheets | d9e064f | 2000-12-13 21:52:37 +0000 | [diff] [blame] | 338 | dsprintf(name, "<message>", ...); |
John R. Sheets | 1e8e5ba | 2000-08-08 01:24:00 +0000 | [diff] [blame] | 339 | </programlisting> |
| 340 | which is just like a <function>sprintf</function> |
| 341 | function but instead of a C string as first parameter it |
| 342 | takes the name you used to declare it. |
| 343 | </para> |
| 344 | </listitem> |
| 345 | <listitem> |
| 346 | <para> |
| 347 | obtain a pointer to the string with: <function>dbg_str(name)</function> |
| 348 | </para> |
| 349 | </listitem> |
| 350 | <listitem> |
| 351 | <para> |
| 352 | reset the string (if you want to reuse it with): |
| 353 | <programlisting> |
| 354 | dbg_reset_str(name); |
| 355 | </programlisting> |
| 356 | </para> |
| 357 | </listitem> |
| 358 | </orderedlist> |
| 359 | |
| 360 | <para> |
| 361 | Example (modified from the code): |
| 362 | </para> |
| 363 | <programlisting> |
| 364 | void some_func(tabs) |
| 365 | { |
| 366 | INT32 i; |
| 367 | LPINT16 p = (LPINT16)tabs; |
| 368 | dbg_decl_str(listbox, 256); /* declare the string */ |
| 369 | |
John R. Sheets | d9e064f | 2000-12-13 21:52:37 +0000 | [diff] [blame] | 370 | for (i = 0; i < descr->nb_tabs; i++) { |
| 371 | descr->tabs[i] = *p++<<1; |
John R. Sheets | 1e8e5ba | 2000-08-08 01:24:00 +0000 | [diff] [blame] | 372 | if(TRACING(listbox)) /* write in it only if |
John R. Sheets | d9e064f | 2000-12-13 21:52:37 +0000 | [diff] [blame] | 373 | dsprintf(listbox, "%hd ", descr->tabs[i]); /* we are gonna output it */ |
John R. Sheets | 1e8e5ba | 2000-08-08 01:24:00 +0000 | [diff] [blame] | 374 | } |
| 375 | TRACE(listbox, "Listbox %04x: settabstops %s", |
John R. Sheets | d9e064f | 2000-12-13 21:52:37 +0000 | [diff] [blame] | 376 | wnd->hwndSelf, dbg_str(listbox)); /* output the whole thing */ |
John R. Sheets | 1e8e5ba | 2000-08-08 01:24:00 +0000 | [diff] [blame] | 377 | } |
| 378 | </programlisting> |
| 379 | <para> |
| 380 | If you need to use it two times in the same scope do like |
| 381 | this: |
| 382 | </para> |
| 383 | <programlisting> |
| 384 | void some_func(tabs) |
| 385 | { |
| 386 | INT32 i; |
| 387 | LPINT16 p = (LPINT16)tabs; |
| 388 | dbg_decl_str(listbox, 256); /* declare the string */ |
| 389 | |
John R. Sheets | d9e064f | 2000-12-13 21:52:37 +0000 | [diff] [blame] | 390 | for (i = 0; i < descr->nb_tabs; i++) { |
| 391 | descr->tabs[i] = *p++<<1; |
John R. Sheets | 1e8e5ba | 2000-08-08 01:24:00 +0000 | [diff] [blame] | 392 | if(TRACING(listbox)) /* write in it only if |
John R. Sheets | d9e064f | 2000-12-13 21:52:37 +0000 | [diff] [blame] | 393 | dsprintf(listbox, "%hd ", descr->tabs[i]); /* we are gonna output it */ |
John R. Sheets | 1e8e5ba | 2000-08-08 01:24:00 +0000 | [diff] [blame] | 394 | } |
| 395 | TRACE(listbox, "Listbox %04x: settabstops %s\n", |
John R. Sheets | d9e064f | 2000-12-13 21:52:37 +0000 | [diff] [blame] | 396 | wnd->hwndSelf, dbg_str(listbox)); /* output the whole thing */ |
John R. Sheets | 1e8e5ba | 2000-08-08 01:24:00 +0000 | [diff] [blame] | 397 | |
| 398 | dbg_reset_str(listbox); /* !!!reset the string!!! */ |
John R. Sheets | d9e064f | 2000-12-13 21:52:37 +0000 | [diff] [blame] | 399 | for (i = 0; i < descr->extrainfo_nr; i++) { |
| 400 | descr->extrainfo = *p+1; |
John R. Sheets | 1e8e5ba | 2000-08-08 01:24:00 +0000 | [diff] [blame] | 401 | if(TRACING(listbox)) /* write in it only if |
John R. Sheets | d9e064f | 2000-12-13 21:52:37 +0000 | [diff] [blame] | 402 | dsprintf(listbox,"%3d ",descr->extrainfo); /* we are gonna output it */ |
John R. Sheets | 1e8e5ba | 2000-08-08 01:24:00 +0000 | [diff] [blame] | 403 | } |
| 404 | |
| 405 | TRACE(listbox, "Listbox %04x: extrainfo %s\n", |
John R. Sheets | d9e064f | 2000-12-13 21:52:37 +0000 | [diff] [blame] | 406 | wnd->hwndSelf, dbg_str(listbox)); /* output the whole thing */ |
John R. Sheets | 1e8e5ba | 2000-08-08 01:24:00 +0000 | [diff] [blame] | 407 | |
| 408 | } |
| 409 | </programlisting> |
| 410 | |
| 411 | <important> |
| 412 | <para> |
| 413 | As I already stated, I do not think this will be the |
| 414 | ultimate interface for building in-memory debugging |
| 415 | messages. In fact, I do have better ideas which I hope to |
| 416 | have time to implement for the next release. For this |
| 417 | reason, please try not to use it. However, if you need to |
| 418 | output a line in more than one |
| 419 | <function>dprintf_xxx</function> calls, then USE THIS |
| 420 | INTERFACE. DO NOT use other methods. This way, I will |
| 421 | easily translate everything to the new interface (when it |
| 422 | will become available). So, if you need to use it, then |
| 423 | follow the following guidelines: |
| 424 | </para> |
| 425 | <itemizedlist> |
| 426 | <listitem> |
| 427 | <para>wrap calls to <function>dsprintf</function> with a |
| 428 | </para> |
| 429 | <programlisting> |
| 430 | if(YYY(xxx)) |
| 431 | dsprintf(xxx,...); |
| 432 | </programlisting> |
| 433 | <para> |
| 434 | Of course, if the call to |
| 435 | <function>dsprintf</function> is made from within a |
| 436 | function which you know is called only if |
| 437 | <function>YYY(xxx)</function> is true, for example if |
| 438 | you call it only like this: |
| 439 | </para> |
| 440 | <programlisting> |
| 441 | if(YYY(xxx)) |
| 442 | print_some_debug_info(); |
| 443 | </programlisting> |
| 444 | <para> |
| 445 | then you need not (and should not) wrap calls to |
| 446 | <function>dsprintf</function> with the before |
| 447 | mentioned <function>if</function>. |
| 448 | </para> |
| 449 | </listitem> |
| 450 | <listitem> |
| 451 | <para> |
| 452 | name the string EXACTLY like the debugging channel on |
| 453 | which is going to be output. Please see the above |
| 454 | example. |
| 455 | </para> |
| 456 | </listitem> |
| 457 | </itemizedlist> |
| 458 | </important> |
John R. Sheets | d9e064f | 2000-12-13 21:52:37 +0000 | [diff] [blame] | 459 | </sect1> |
John R. Sheets | 1e8e5ba | 2000-08-08 01:24:00 +0000 | [diff] [blame] | 460 | |
John R. Sheets | d9e064f | 2000-12-13 21:52:37 +0000 | [diff] [blame] | 461 | <sect1 id="dbg-resource-ids"> |
John R. Sheets | 1e8e5ba | 2000-08-08 01:24:00 +0000 | [diff] [blame] | 462 | <title>Resource identifiers</title> |
| 463 | |
| 464 | <para> |
| 465 | Resource identifiers can be either strings or numbers. To |
| 466 | make life a bit easier for outputting these beasts (and to |
| 467 | help you avoid the need to build the message in memory), I |
| 468 | introduced a new function called <function>debugres</function>. |
| 469 | </para> |
| 470 | <para> |
| 471 | The function is defined in <filename>debugstr.h</filename> |
| 472 | and has the following prototype: |
| 473 | </para> |
| 474 | <programlisting> |
| 475 | LPSTR debugres(const void *id); |
| 476 | </programlisting> |
| 477 | <para> |
| 478 | It takes a pointer to the resource id and returns a nicely |
| 479 | formatted string of the identifier. If the high word of the |
| 480 | pointer is <literal>0</literal>, then it assumes that the |
| 481 | identifier is a number and thus returns a string of the |
| 482 | form: |
| 483 | </para> |
| 484 | <programlisting> |
| 485 | #xxxx |
| 486 | </programlisting> |
| 487 | <para> |
| 488 | where <literal>xxxx</literal> are 4 hex-digits representing |
| 489 | the low word of <parameter>id</parameter>. |
| 490 | </para> |
| 491 | <para> |
| 492 | If the high word of the pointer is not <literal>0</literal>, |
| 493 | then it assumes that the identifier is a string and thus |
| 494 | returns a string of the form: |
| 495 | </para> |
| 496 | <programlisting> |
John R. Sheets | d9e064f | 2000-12-13 21:52:37 +0000 | [diff] [blame] | 497 | '<identifier>' |
John R. Sheets | 1e8e5ba | 2000-08-08 01:24:00 +0000 | [diff] [blame] | 498 | </programlisting> |
| 499 | <para> |
| 500 | Thus, to use it, do something on the following lines: |
| 501 | </para> |
| 502 | <programlisting> |
| 503 | #include "debug.h" |
| 504 | |
| 505 | ... |
| 506 | |
| 507 | YYY(xxx, "resource is %s", debugres(myresource)); |
| 508 | </programlisting> |
John R. Sheets | d9e064f | 2000-12-13 21:52:37 +0000 | [diff] [blame] | 509 | </sect1> |
John R. Sheets | 1e8e5ba | 2000-08-08 01:24:00 +0000 | [diff] [blame] | 510 | |
John R. Sheets | d9e064f | 2000-12-13 21:52:37 +0000 | [diff] [blame] | 511 | <sect1 id="dbg-param"> |
John R. Sheets | 1e8e5ba | 2000-08-08 01:24:00 +0000 | [diff] [blame] | 512 | <title>The <parameter>--debugmsg</parameter> command line option</title> |
| 513 | |
| 514 | <para> |
| 515 | So, the <parameter>--debugmsg</parameter> command line |
| 516 | option has been changed as follows: |
| 517 | </para> |
| 518 | <itemizedlist> |
| 519 | <listitem> |
| 520 | <para> |
| 521 | the new syntax is: <parameter>--debugmsg |
| 522 | [yyy]#xxx[,[yyy1]#xxx1]*</parameter> where |
| 523 | <literal>#</literal> is either <literal>+</literal> or |
| 524 | <literal>-</literal> |
| 525 | </para> |
| 526 | </listitem> |
| 527 | <listitem> |
| 528 | <para> |
| 529 | when the optional class argument (<literal>yyy</literal>) |
| 530 | is not present, then the statement will |
| 531 | enable(<literal>+</literal>)/disable(<literal>-</literal>) |
| 532 | all messages for the given channel (<literal>xxx</literal>) |
| 533 | on all classes. For example: |
| 534 | </para> |
| 535 | <programlisting> |
| 536 | --debugmsg +reg,-file |
| 537 | </programlisting> |
| 538 | <para> |
| 539 | enables all messages on the <literal>reg</literal> |
| 540 | channel and disables all messages on the |
| 541 | <literal>file</literal> channel. This is same as the old |
| 542 | semantics. |
| 543 | </para> |
| 544 | </listitem> |
| 545 | <listitem> |
| 546 | <para> |
| 547 | when the optional class argument (<literal>yyy</literal>) |
| 548 | is present, then the statement will enable |
| 549 | (<literal>+</literal>)/disable(<literal>-</literal>) |
| 550 | messages for the given channel (<literal>xxx</literal>) |
| 551 | only on the given class. For example: |
| 552 | </para> |
| 553 | <programlisting> |
| 554 | --debugmsg trace+reg,warn-file |
| 555 | </programlisting> |
| 556 | <para> |
| 557 | enables trace messages on the <literal>reg</literal> |
| 558 | channel and disables warning messages on the |
| 559 | <literal>file</literal> channel. |
| 560 | </para> |
| 561 | </listitem> |
| 562 | <listitem> |
| 563 | <para> |
| 564 | also, the pseudo-channel all is also supported and it |
| 565 | has the intuitive semantics: |
| 566 | </para> |
| 567 | <screen> |
| 568 | --debugmsg +all -- enables all debug messages |
| 569 | --debugmsg -all -- disables all debug messages |
| 570 | --debugmsg yyy+all -- enables debug messages for class yyy on all |
| 571 | channels. |
| 572 | --debugmsg yyy-all -- disables debug messages for class yyy on all |
| 573 | channels. |
| 574 | </screen> |
| 575 | <para> |
| 576 | So, for example: |
| 577 | </para> |
| 578 | <screen> |
| 579 | --debugmsg warn-all -- disables all warning messages. |
| 580 | </screen> |
| 581 | </listitem> |
| 582 | </itemizedlist> |
| 583 | |
| 584 | <para> |
| 585 | Also, note that at the moment: |
| 586 | </para> |
| 587 | <itemizedlist> |
| 588 | <listitem> |
| 589 | <para>the <literal>FIXME</literal> and <literal>ERR</literal> |
| 590 | classes are enabled by default</para> |
| 591 | </listitem> |
| 592 | <listitem> |
| 593 | <para>the <literal>TRACE</literal> and |
| 594 | <literal>WARN</literal> classes are disabled by |
| 595 | default</para> |
| 596 | </listitem> |
| 597 | </itemizedlist> |
John R. Sheets | d9e064f | 2000-12-13 21:52:37 +0000 | [diff] [blame] | 598 | </sect1> |
John R. Sheets | 1e8e5ba | 2000-08-08 01:24:00 +0000 | [diff] [blame] | 599 | |
John R. Sheets | d9e064f | 2000-12-13 21:52:37 +0000 | [diff] [blame] | 600 | <sect1 id="dbg-compiling"> |
John R. Sheets | 1e8e5ba | 2000-08-08 01:24:00 +0000 | [diff] [blame] | 601 | <title>Compiling Out Debugging Messages</title> |
| 602 | |
| 603 | <para> |
| 604 | To compile out the debugging messages, provide |
| 605 | <command>configure</command> with the following options: |
| 606 | </para> |
| 607 | <screen> |
| 608 | --disable-debug -- turns off TRACE, WARN, and FIXME (and DUMP). |
| 609 | --disable-trace -- turns off TRACE only. |
| 610 | </screen> |
| 611 | <para> |
| 612 | This will result in an executable that, when stripped, is |
| 613 | about 15%-20% smaller. Note, however, that you will not be |
| 614 | able to effectively debug Wine without these messages. |
| 615 | </para> |
| 616 | <para> |
| 617 | This feature has not been extensively tested--it may subtly |
| 618 | break some things. |
| 619 | </para> |
John R. Sheets | d9e064f | 2000-12-13 21:52:37 +0000 | [diff] [blame] | 620 | </sect1> |
John R. Sheets | 1e8e5ba | 2000-08-08 01:24:00 +0000 | [diff] [blame] | 621 | |
John R. Sheets | d9e064f | 2000-12-13 21:52:37 +0000 | [diff] [blame] | 622 | <sect1 id="dbg-notes"> |
John R. Sheets | 1e8e5ba | 2000-08-08 01:24:00 +0000 | [diff] [blame] | 623 | <title>A Few Notes on Style</title> |
| 624 | |
| 625 | <para> |
| 626 | This new scheme makes certain things more consistent but |
| 627 | there is still room for improvement by using a common style |
| 628 | of debug messages. Before I continue, let me note that the |
| 629 | output format is the following: |
| 630 | </para> |
| 631 | <screen> |
John R. Sheets | d9e064f | 2000-12-13 21:52:37 +0000 | [diff] [blame] | 632 | yyy:xxx:fff <message> |
John R. Sheets | 1e8e5ba | 2000-08-08 01:24:00 +0000 | [diff] [blame] | 633 | |
| 634 | where: |
| 635 | yyy = the class (fixme, err, warn, trace) |
| 636 | xxx = the channel (atom, win, font, etc) |
| 637 | fff = the function name |
| 638 | </screen> |
| 639 | <para> |
| 640 | these fields are output automatically. All you have to |
John R. Sheets | d9e064f | 2000-12-13 21:52:37 +0000 | [diff] [blame] | 641 | provide is the <message> part. |
John R. Sheets | 1e8e5ba | 2000-08-08 01:24:00 +0000 | [diff] [blame] | 642 | </para> |
| 643 | <para> |
| 644 | So here are some ideas: |
| 645 | </para> |
| 646 | |
| 647 | <itemizedlist> |
| 648 | <listitem> |
| 649 | <para>do NOT include the name of the function: it is included automatically</para> |
| 650 | </listitem> |
| 651 | <listitem> |
| 652 | <para> |
| 653 | if you want to output the parameters of the function, do |
| 654 | it as the first thing and include them in parentheses, |
| 655 | like this: |
| 656 | <programlisting> |
| 657 | YYY(xxx, "(%d,%p,etc)...\n", par1, par2, ...); |
| 658 | </programlisting> |
| 659 | </para> |
| 660 | </listitem> |
| 661 | <listitem> |
| 662 | <para> |
| 663 | for stubs, you should output a <literal>FIXME</literal> |
| 664 | message. I suggest this style: |
| 665 | <programlisting> |
| 666 | FIXME(xxx, "(%x,%d...): stub\n", par1, par2, ...); |
| 667 | </programlisting> |
| 668 | That is, you output the parameters, then a : and then a string |
| 669 | containing the word "stub". I've seen "empty stub", and others, but I |
| 670 | think that just "stub" suffices. |
| 671 | </para> |
| 672 | </listitem> |
| 673 | <listitem> |
| 674 | <para> |
| 675 | output 1 and ONLY 1 line per message. That is, the format |
| 676 | string should contain only 1 <literal>\n</literal> and it |
| 677 | should always appear at the end of the string. (there are |
| 678 | many reasons for this requirement, one of them is that |
| 679 | each debug macro adds things to the beginning of the line) |
| 680 | </para> |
| 681 | </listitem> |
| 682 | <listitem> |
| 683 | <para> |
| 684 | if you want to name a value, use <literal>=</literal> and |
| 685 | NOT <literal>:</literal>. That is, instead of saying: |
| 686 | <programlisting> |
| 687 | FIXME(xxx, "(fd: %d, file: %s): stub\n", fd, name); |
| 688 | </programlisting> |
| 689 | say: |
| 690 | <programlisting> |
| 691 | FIXME(xxx, "(fd=%d, file=%s): stub\n", fd, name); |
| 692 | </programlisting> |
| 693 | use <literal>:</literal> to separate categories. |
| 694 | </para> |
| 695 | </listitem> |
| 696 | <listitem> |
| 697 | <para> |
| 698 | try to avoid the style: |
| 699 | <programlisting> |
Simeon Pilgrim | d5518e7 | 2001-05-02 01:10:16 +0000 | [diff] [blame] | 700 | FIXME(xxx, "(fd=%d, file=%s)\n", fd, name); |
John R. Sheets | 1e8e5ba | 2000-08-08 01:24:00 +0000 | [diff] [blame] | 701 | </programlisting> |
| 702 | but use: |
| 703 | <programlisting> |
| 704 | FIXME(xxx, "(fd=%d, file=%s): stub\n", fd, name); |
| 705 | </programlisting> |
| 706 | The reason is that if you want to <command>grep</command> |
| 707 | for things, you would search for <literal>FIXME</literal> |
| 708 | but in the first case there is no additional information |
| 709 | available, where in the second one, there is (e.g. the word |
| 710 | stub) |
| 711 | </para> |
| 712 | </listitem> |
| 713 | <listitem> |
| 714 | <para> |
| 715 | if you output a string s that might contain control |
| 716 | characters, or if <parameter>s</parameter> may be |
| 717 | <literal>NULL</literal>, use |
| 718 | <function>debugstr_a</function> (for ASCII strings, or |
| 719 | <function>debugstr_w</function> for Unicode strings) to |
| 720 | convert <parameter>s</parameter> to a C string, like this: |
| 721 | <programlisting> |
| 722 | HANDLE32 WINAPI YourFunc(LPCSTR s) |
| 723 | { |
| 724 | FIXME(xxx, "(%s): stub\n", debugstr_a(s)); |
| 725 | } |
| 726 | </programlisting> |
| 727 | </para> |
| 728 | </listitem> |
| 729 | <listitem> |
| 730 | <para> |
| 731 | if you want to output a resource identifier, use debugres to |
| 732 | convert it to a string first, like this: |
| 733 | <programlisting> |
| 734 | HANDLE32 WINAPI YourFunc(LPCSTR res) |
| 735 | { |
| 736 | FIXME(xxx, "(res=%s): stub\n", debugres(s)); |
| 737 | } |
| 738 | </programlisting> |
| 739 | if the resource identifier is a <type>SEGPTR</type>, use |
| 740 | <function>PTR_SEG_TO_LIN</function> to get a |
| 741 | liner pointer first: |
| 742 | <programlisting> |
| 743 | HRSRC16 WINAPI FindResource16( HMODULE16 hModule, SEGPTR name, SEGPTR type ) |
| 744 | { |
| 745 | [...] |
| 746 | TRACE(resource, "module=%04x name=%s type=%s\n", |
| 747 | hModule, debugres(PTR_SEG_TO_LIN(name)), |
| 748 | debugres(PTR_SEG_TO_LIN(type)) ); |
| 749 | [...] |
| 750 | } |
| 751 | </programlisting> |
| 752 | </para> |
| 753 | </listitem> |
| 754 | <listitem> |
| 755 | <para> |
| 756 | for messages intended for the user (specifically those that |
| 757 | report errors in <filename>wine.conf</filename>), use the |
| 758 | <literal>MSG</literal> macro. Use it like a |
| 759 | <function>printf</function>: |
| 760 | <programlisting> |
| 761 | MSG( "Definition of drive %d is incorrect!\n", drive ); |
| 762 | </programlisting> |
| 763 | However, note that there are <emphasis>very</emphasis> few |
| 764 | valid uses of this macro. Most messages are debugging |
| 765 | messages, so chances are you will not need to use this |
| 766 | macro. Grep the source to get an idea where it is |
| 767 | appropriate to use it. |
| 768 | </para> |
| 769 | </listitem> |
| 770 | <listitem> |
| 771 | <para> |
| 772 | For structure dumps, use the <function>DUMP</function> |
| 773 | macro. Use it like a <function>printf</function>, just like |
| 774 | the <literal>MSG</literal> macro. Similarly, there are only |
| 775 | a few valid uses of this macro. Grep the source to see when |
| 776 | to use it. |
| 777 | </para> |
| 778 | </listitem> |
| 779 | </itemizedlist> |
John R. Sheets | d9e064f | 2000-12-13 21:52:37 +0000 | [diff] [blame] | 780 | </sect1> |
John R. Sheets | 1e8e5ba | 2000-08-08 01:24:00 +0000 | [diff] [blame] | 781 | |
John R. Sheets | 1e8e5ba | 2000-08-08 01:24:00 +0000 | [diff] [blame] | 782 | </chapter> |
| 783 | |
| 784 | <!-- Keep this comment at the end of the file |
| 785 | Local variables: |
| 786 | mode: sgml |
John R. Sheets | d9e064f | 2000-12-13 21:52:37 +0000 | [diff] [blame] | 787 | sgml-parent-document:("wine-doc.sgml" "set" "book" "part" "chapter" "") |
John R. Sheets | 1e8e5ba | 2000-08-08 01:24:00 +0000 | [diff] [blame] | 788 | End: |
| 789 | --> |