]>
Commit | Line | Data |
---|---|---|
f2d76545 | 1 | Noteworthy changes in GCC for EGCS. |
fc5a8790 | 2 | |
e5e809f4 JL |
3 | Noteworthy changes in GCC version 2.8.1 |
4 | --------------------------------------- | |
5 | ||
6 | Numerous bugs have been fixed and some minor performance | |
7 | improvements (compilation speed) have been made. | |
8 | ||
9 | Noteworthy changes in GCC version 2.8.0 | |
10 | --------------------------------------- | |
11 | ||
956d6950 JL |
12 | A major change in this release is the addition of a framework for |
13 | exception handling, currently used by C++. Many internal changes and | |
14 | optimization improvements have been made. These increase the | |
15 | maintainability and portability of GCC. GCC now uses autoconf to | |
16 | compute many host parameters. | |
c6258ee2 | 17 | |
956d6950 | 18 | The following lists changes that add new features or targets. |
861bb6c1 | 19 | |
956d6950 | 20 | See cp/NEWS for new features of C++ in this release. |
fc5a8790 | 21 | |
956d6950 | 22 | New tools and features: |
fc5a8790 | 23 | |
956d6950 JL |
24 | The Dwarf 2 debugging information format is supported on ELF systems, and |
25 | is the default for -ggdb where available. It can also be used for C++. | |
26 | The Dwarf version 1 debugging format is also permitted for C++, but | |
27 | does not work well. | |
fc5a8790 | 28 | |
956d6950 JL |
29 | gcov.c is provided for test coverage analysis and branch profiling |
30 | analysis is also supported; see -fprofile-arcs, -ftest-coverage, | |
31 | and -fbranch-probabilities. | |
32 | ||
33 | Support for the Checker memory checking tool. | |
34 | ||
35 | New switch, -fstack-check, to check for stack overflow on systems that | |
36 | don't have such built into their ABI. | |
37 | ||
38 | New switches, -Wundef and -Wno-undef to warn if an undefined identifier | |
39 | is evaluated in an #if directive. | |
40 | ||
41 | Options -Wall and -Wimplicit now cause GCC to warn about implicit int | |
42 | in declarations (e.g. `register i;'), since the C Standard committee | |
43 | has decided to disallow this in the next revision of the standard; | |
44 | -Wimplicit-function-declarations and -Wimplicit-int are subsets of | |
45 | this. | |
46 | ||
47 | Option -Wsign-compare causes GCC to warn about comparison of signed and | |
48 | unsigned values. | |
49 | ||
50 | Add -dI option of cccp for cxref. | |
51 | ||
52 | New features in configuration, installation and specs file handling: | |
53 | ||
54 | New option --enable-c-cpplib to configure script. | |
55 | ||
b472527b JL |
56 | You can use --with-cpu on the configure command to specify the default |
57 | CPU that GCC should generate code for. | |
956d6950 JL |
58 | |
59 | The -specs=file switch allows you to override default specs used in | |
60 | invoking programs like cc1, as, etc. | |
61 | ||
62 | Allow including one specs file from another and renaming a specs | |
63 | variable. | |
64 | ||
65 | You can now relocate all GCC files with a single environment variable | |
66 | or a registry entry under Windows 95 and Windows NT. | |
67 | ||
68 | Changes in Objective-C: | |
69 | ||
70 | The Objective-C Runtime Library has been made thread-safe. | |
71 | ||
72 | The Objective-C Runtime Library contains an interface for creating | |
73 | mutexes, condition mutexes, and threads; it requires a back-end | |
74 | implementation for the specific platform and/or thread package. | |
75 | Currently supported are DEC/OSF1, IRIX, Mach, OS/2, POSIX, PCThreads, | |
76 | Solaris, and Windows32. The --enable-threads parameter can be used | |
77 | when configuring GCC to enable and select a thread back-end. | |
78 | ||
79 | Objective-C is now configured as separate front-end language to GCC, | |
80 | making it more convenient to conditionally build it. | |
fc5a8790 | 81 | |
956d6950 JL |
82 | The internal structures of the Objective-C Runtime Library have |
83 | changed sufficiently to warrant a new version number; now version 8. | |
84 | Programs compiled with an older version must be recompiled. | |
85 | ||
86 | The Objective-C Runtime Library can be built as a DLL on Windows 95 | |
87 | and Windows NT systems. | |
88 | ||
89 | The Objective-C Runtime Library implements +load. | |
90 | ||
91 | The following new targets are supported (see also list under each | |
92 | individual CPU below): | |
93 | ||
94 | Embedded target m32r-elf. | |
95 | Embedded Hitachi Super-H using ELF. | |
96 | RTEMS real-time system on various CPU targets. | |
97 | ARC processor. | |
98 | NEC V850 processor. | |
99 | Matsushita MN10200 processor. | |
100 | Matsushita MN10300 processor. | |
101 | Sparc and PowerPC running on VxWorks. | |
102 | Support both glibc versions 1 and 2 on Linux-based GNU systems. | |
103 | ||
104 | New features for DEC Alpha systems: | |
105 | ||
106 | Allow detailed specification of IEEE fp support: | |
107 | -mieee, -mieee-with-inexact, and -mieee-conformant | |
108 | -mfp-trap-mode=xxx, -mfp-round-mode=xxx, -mtrap-precision=xxx | |
109 | -mcpu=xxx for CPU selection | |
110 | Support scheduling parameters for EV5. | |
111 | Add support for BWX, CIX, and MAX instruction set extensions. | |
112 | Support Linux-based GNU systems. | |
113 | Support VMS. | |
114 | ||
115 | Additional supported processors and systems for MIPS targets: | |
116 | ||
117 | MIPS4 instruction set. | |
118 | R4100, R4300 and R5000 processors. | |
119 | N32 and N64 ABI. | |
120 | IRIX 6.2. | |
121 | SNI SINIX. | |
122 | ||
123 | New features for Intel x86 family: | |
124 | ||
125 | Add scheduling parameters for Pentium and Pentium Pro. | |
126 | Support stabs on Solaris-x86. | |
127 | Intel x86 processors running the SCO OpenServer 5 family. | |
128 | Intel x86 processors running DG/UX. | |
129 | Intel x86 using Cygwin32 or Mingw32 on Windows 95 and Windows NT. | |
130 | ||
131 | New features for Motorola 68k family: | |
132 | ||
133 | Support for 68060 processor. | |
134 | More consistent switches to specify processor. | |
135 | Motorola 68k family running AUX. | |
136 | 68040 running pSOS, ELF object files, DBX debugging. | |
137 | Coldfire variant of Motorola m68k family. | |
138 | ||
139 | New features for the HP PA RISC: | |
140 | ||
141 | -mspace and m-no-space | |
142 | -mlong-load-store and -mno-long-load-store | |
143 | -mbig-switch -mno-big-switch | |
144 | ||
145 | GCC on the PA requires either gas-2.7 or the HP assembler; for best | |
146 | results using GAS is highly recommended. GAS is required for -g and | |
147 | exception handling support. | |
fc5a8790 JL |
148 | |
149 | New features for SPARC-based systems: | |
150 | ||
956d6950 JL |
151 | The ultrasparc cpu. |
152 | The sparclet cpu, supporting only a.out file format. | |
153 | Sparc running SunOS 4 with the GNU assembler. | |
154 | Sparc running the Linux-based GNU system. | |
155 | Embedded Sparc processors running the ELF object file format. | |
fc5a8790 JL |
156 | -mcpu=xxx |
157 | -mtune=xxx | |
f2d76545 JL |
158 | -malign-loops=xxx |
159 | -malign-jumps=xxx | |
160 | -malign-functions=xxx | |
fc5a8790 JL |
161 | -mimpure-text and -mno-impure-text |
162 | ||
956d6950 JL |
163 | Options -mno-v8 and -mno-sparclite are no longer supported on SPARC |
164 | targets. Options -mcypress, -mv8, -msupersparc, -msparclite, -mf930, | |
165 | and -mf934 are deprecated and will be deleted in GCC 2.9. Use | |
166 | -mcpu=xxx instead. | |
861bb6c1 | 167 | |
956d6950 | 168 | New features for rs6000 and PowerPC systems: |
861bb6c1 | 169 | |
956d6950 JL |
170 | Solaris 2.51 running on PowerPC's. |
171 | The Linux-based GNU system running on PowerPC's. | |
172 | -mcpu=604e,602,603e,620,801,823,mpc505,821,860,power2 | |
173 | -mtune=xxx | |
174 | -mrelocatable-lib, m-no-relocatable-lib | |
175 | -msim, -mmve, -memb | |
176 | -mupdate, -mno-update | |
177 | -mfused-madd, -mno-fused-madd | |
fc5a8790 JL |
178 | |
179 | -mregnames | |
180 | -meabi | |
956d6950 JL |
181 | -mcall-linux, -mcall-solaris, -mcall-sysv-eabi, -mcall-sysv-noeabi |
182 | -msdata, -msdata=none, -msdata=default, -msdata=sysv, -msdata=eabi | |
183 | -memb, -msim, -mmvme | |
184 | -myellowknife, -mads | |
185 | wchar_t is now of type long as per the ABI, not unsigned short. | |
861bb6c1 JL |
186 | -p/-pg support |
187 | -mcpu=403 now implies -mstrict-align. | |
956d6950 JL |
188 | Implement System V profiling. |
189 | ||
190 | Aix 4.1 GCC targets now default to -mcpu=common so that programs | |
191 | compiled can be moved between rs6000 and powerpc based systems. A | |
192 | consequence of this is that -static won't work, and that some programs | |
193 | may be slightly slower. | |
fc5a8790 | 194 | |
956d6950 JL |
195 | You can select the default value to use for -mcpu=xxx on rs6000 and |
196 | powerpc targets by using the --with-cpu=xxx option when configuring the | |
197 | compiler. In addition, a new options, -mtune=xxx was added that | |
198 | selects the machine to schedule for but does not select the | |
199 | architecture level. | |
fc5a8790 | 200 | |
956d6950 JL |
201 | Directory names used for storing the multilib libraries on System V |
202 | and embedded PowerPC systems have been shortened to work with commands | |
203 | like tar that have fixed limits on pathname size. | |
fc5a8790 | 204 | |
956d6950 JL |
205 | New features for the Hitachi H8/300(H): |
206 | ||
207 | -malign-300 | |
208 | -ms (for the Hitachi H8/S processor) | |
209 | -mint32 | |
fc5a8790 | 210 | |
956d6950 | 211 | New features for the ARM: |
fc5a8790 | 212 | |
956d6950 JL |
213 | -march=xxx, -mtune=xxx, -mcpu=xxx |
214 | Support interworking with Thumb code. | |
215 | ARM processor with a.out object format, COFF, or AOF assembler. | |
216 | ARM on "semi-hosted" platform. | |
217 | ARM running NetBSD. | |
218 | ARM running the Linux-based GNU system. | |
fc5a8790 | 219 | |
956d6950 | 220 | New feature for Solaris systems: |
f2d76545 | 221 | |
956d6950 JL |
222 | GCC installation no longer makes a copy of system include files, |
223 | thus insulating GCC better from updates to the operating system. | |
224 | ||
956d6950 JL |
225 | \f |
226 | Noteworthy changes in GCC version 2.7.2 | |
227 | --------------------------------------- | |
fc5a8790 JL |
228 | |
229 | A few bugs have been fixed (most notably the generation of an | |
230 | invalid assembler opcode on some RS/6000 systems). | |
231 | ||
956d6950 JL |
232 | Noteworthy changes in GCC version 2.7.1 |
233 | --------------------------------------- | |
fc5a8790 JL |
234 | |
235 | This release fixes numerous bugs (mostly minor) in GCC 2.7.0, but | |
236 | also contains a few new features, mostly related to specific targets. | |
237 | ||
238 | Major changes have been made in code to support Windows NT. | |
239 | ||
240 | The following new targets are supported: | |
241 | ||
242 | 2.9 BSD on PDP-11 | |
243 | Linux on m68k | |
244 | HP/UX version 10 on HP PA RISC (treated like version 9) | |
245 | DEC Alpha running Windows NT | |
246 | ||
247 | When parsing C, GCC now recognizes C++ style `//' comments unless you | |
248 | specify `-ansi' or `-traditional'. | |
249 | ||
250 | The PowerPC System V targets (powerpc-*-sysv, powerpc-*-eabi) now use the | |
251 | calling sequence specified in the System V Application Binary Interface | |
252 | Processor Supplement (PowerPC Processor ABI Supplement) rather than the calling | |
253 | sequence used in GCC version 2.7.0. That calling sequence was based on the AIX | |
254 | calling sequence without function descriptors. To compile code for that older | |
255 | calling sequence, either configure the compiler for powerpc-*-eabiaix or use | |
256 | the -mcall-aix switch when compiling and linking. | |
956d6950 JL |
257 | \f |
258 | Noteworthy changes in GCC version 2.7.0 | |
259 | --------------------------------------- | |
fc5a8790 JL |
260 | |
261 | GCC now works better on systems that use ".obj" and ".exe" instead of | |
262 | ".o" and no extension. This involved changes to the driver program, | |
263 | gcc.c, to convert ".o" names to ".obj" and to GCC's Makefile to use | |
264 | ".obj" and ".exe" in filenames that are not targets. In order to | |
265 | build GCC on such systems, you may need versions of GNU make and/or | |
266 | compatible shells. At this point, this support is preliminary. | |
267 | ||
268 | Object file extensions of ".obj" and executable file extensions of | |
269 | ".exe" are allowed when using appropriate version of GNU Make. | |
270 | ||
271 | Numerous enhancements were made to the __attribute__ facility including | |
272 | more attributes and more places that support it. We now support the | |
273 | "packed", "nocommon", "noreturn", "volatile", "const", "unused", | |
274 | "transparent_union", "constructor", "destructor", "mode", "section", | |
275 | "align", "format", "weak", and "alias" attributes. Each of these | |
276 | names may also be specified with added underscores, e.g., "__packed__". | |
277 | __attribute__ may now be applied to parameter definitions, function | |
278 | definitions, and structure, enum, and union definitions. | |
279 | ||
280 | GCC now supports returning more structures in registers, as specified by | |
281 | many calling sequences (ABIs), such as on the HP PA RISC. | |
282 | ||
283 | A new option '-fpack-struct' was added to automatically pack all structure | |
284 | members together without holes. | |
285 | ||
286 | There is a new library (cpplib) and program (cppmain) that at some | |
287 | point will replace cpp (aka cccp). To use cppmain as cpp now, pass | |
288 | the option CCCP=cppmain to make. The library is already used by the | |
289 | fix-header program, which should speed up the fixproto script. | |
290 | ||
291 | New options for supported targets: | |
292 | ||
293 | GNU on many targets. | |
294 | NetBSD on MIPS, m68k, VAX, and x86. | |
295 | LynxOS on x86, m68k, Sparc, and RS/6000. | |
296 | VxWorks on many targets. | |
297 | ||
298 | Windows/NT on x86 architecture. Initial support for Windows/NT on Alpha | |
299 | (not fully working). | |
300 | ||
301 | Many embedded targets, specifically UDI on a29k, aout, coff, elf, | |
302 | and vsta "operating systems" on m68k, m88k, mips, sparc, and x86. | |
303 | ||
304 | Additional support for x86 (i386, i486, and Pentium): | |
305 | ||
306 | Work with old and new linkers for Linux-based GNU systems, | |
307 | supporting both a.out and ELF. | |
308 | FreeBSD on x86. | |
309 | Stdcall convention. | |
310 | -malign-double, -mregparm=, -malign-loops= and -malign-jumps= switches. | |
311 | On ISC systems, support -Xp like -posix. | |
312 | ||
313 | Additions for RS/6000: | |
314 | ||
315 | Instruction scheduling information for PowerPC 403. | |
316 | AIX 4.1 on PowerPC. | |
317 | -mstring and -mno-string. | |
318 | -msoft-float and floating-point emulation included. | |
319 | Preliminary support for PowerPC System V.4 with or without the GNU as. | |
320 | Preliminary support for EABI. | |
321 | Preliminary support for 64-bit systems. | |
322 | Both big and little endian systems. | |
323 | ||
324 | New features for MIPS-based systems: | |
325 | ||
326 | r4650. | |
327 | mips4 and R8000. | |
328 | Irix 6.0. | |
329 | 64-bit ABI. | |
330 | Allow dollar signs in labels on SGI/Irix 5.x. | |
331 | ||
332 | New support for HP PA RISC: | |
333 | ||
334 | Generation of PIC (requires binutils-2.5.2.u6 or later). | |
335 | HP-UX version 9 on HP PA RISC (dynamically links even with -g). | |
336 | Processor variants for HP PA RISC: 700, 7100, and 7100LC. | |
337 | Automatic generation of long calls when needed. | |
338 | -mfast-indirect-calls for kernels and static binaries. | |
339 | ||
340 | The called routine now copies arguments passed by invisible reference, | |
341 | as required by the calling standard. | |
342 | ||
343 | Other new miscellaneous target-specific support: | |
344 | ||
345 | -mno-multm on a29k. | |
346 | -mold-align for i960. | |
347 | Configuration for "semi-hosted" ARM. | |
348 | -momit-leaf-frame-pointer for M88k. | |
349 | SH3 variant of Hitachi Super-H and support both big and little endian. | |
350 | ||
351 | Changes to Objective-C: | |
352 | ||
353 | Bare-bones implementation of NXConstantString has been added, | |
354 | which is invoked by the @"string" directive. | |
355 | ||
356 | Class * has been changed to Class to conform to the NextSTEP and | |
357 | OpenStep runtime. | |
358 | ||
359 | Enhancements to make dynamic loading easier. | |
360 | ||
361 | The module version number has been updated to Version 7, thus existing | |
362 | code will need to be recompiled to use the current run-time library. | |
363 | ||
364 | GCC now supports the ISO Normative Addendum 1 to the C Standard. | |
365 | As a result: | |
366 | ||
367 | The header <iso646.h> defines macros for C programs written | |
368 | in national variants of ISO 646. | |
369 | ||
370 | The following digraph tokens are supported: | |
371 | <: :> <% %> %: %:%: | |
372 | These behave like the following, respectively: | |
373 | [ ] { } # ## | |
374 | ||
375 | Digraph tokens are supported unless you specify the `-traditional' | |
376 | option; you do not need to specify `-ansi' or `-trigraphs'. Except | |
377 | for contrived and unlikely examples involving preprocessor | |
378 | stringizing, digraph interpretation doesn't change the meaning of | |
379 | programs; this is unlike trigraph interpretation, which changes the | |
380 | meanings of relatively common strings. | |
381 | ||
382 | The macro __STDC_VERSION__ has the value 199409L. | |
383 | ||
384 | As usual, for full conformance to the standard, you also need a | |
385 | C library that conforms. | |
386 | ||
387 | The following lists changes that have been made to g++. If some | |
388 | features mentioned below sound unfamiliar, you will probably want to | |
389 | look at the recently-released public review copy of the C++ Working | |
390 | Paper. For PostScript and PDF (Adobe Acrobat) versions, see the | |
391 | archive at ftp://research.att.com/dist/stdc++/WP. For HTML and ASCII | |
392 | versions, see ftp://ftp.cygnus.com/pub/g++. On the web, see | |
393 | http://www.cygnus.com/~mrs/wp-draft. | |
394 | ||
395 | The scope of variables declared in the for-init-statement has been changed | |
396 | to conform to http://www.cygnus.com/~mrs/wp-draft/stmt.html#stmt.for; as a | |
397 | result, packages such as groff 1.09 will not compile unless you specify the | |
398 | -fno-for-scope flag. PLEASE DO NOT REPORT THIS AS A BUG; this is a change | |
399 | mandated by the C++ standardization committee. | |
400 | ||
401 | Binary incompatibilities: | |
402 | ||
403 | The builtin 'bool' type is now the size of a machine word on RISC targets, | |
404 | for code efficiency; it remains one byte long on CISC targets. | |
405 | ||
406 | Code that does not use #pragma interface/implementation will most | |
407 | likely shrink dramatically, as g++ now only emits the vtable for a | |
408 | class in the translation unit where its first non-inline, non-abstract | |
409 | virtual function is defined. | |
410 | ||
411 | Classes that do not define the copy constructor will sometimes be | |
412 | passed and returned in registers. This may illuminate latent bugs in | |
413 | your code. | |
414 | ||
415 | Support for automatic template instantiation has *NOT* been added, due | |
416 | to a disagreement over design philosophies. | |
417 | ||
418 | Support for exception handling has been improved; more targets are now | |
419 | supported, and throws will use the RTTI mechanism to match against the | |
420 | catch parameter type. Optimization is NOT SUPPORTED with | |
421 | -fhandle-exceptions; no need to report this as a bug. | |
422 | ||
423 | Support for Run-Time Type Identification has been added with -frtti. | |
424 | This support is still in alpha; one major restriction is that any file | |
425 | compiled with -frtti must include <typeinfo.h>. | |
426 | ||
427 | Preliminary support for namespaces has been added. This support is far | |
428 | from complete, and probably not useful. | |
429 | ||
430 | Synthesis of compiler-generated constructors, destructors and | |
431 | assignment operators is now deferred until the functions are used. | |
432 | ||
433 | The parsing of expressions such as `a ? b : c = 1' has changed from | |
434 | `(a ? b : c) = 1' to `a : b ? (c = 1)'. | |
435 | ||
436 | The code generated for testing conditions, especially those using || | |
437 | and &&, is now more efficient. | |
438 | ||
439 | The operator keywords and, and_eq, bitand, bitor, compl, not, not_eq, | |
440 | or, or_eq, xor and xor_eq are now supported. Use -ansi or | |
441 | -foperator-names to enable them. | |
442 | ||
443 | The 'explicit' keyword is now supported. 'explicit' is used to mark | |
444 | constructors and type conversion operators that should not be used | |
445 | implicitly. | |
446 | ||
447 | g++ now accepts the typename keyword, though it currently has no | |
448 | semantics; it can be a no-op in the current template implementation. | |
449 | You may want to start using it in your code, however, since the | |
450 | pending rewrite of the template implementation to compile STL properly | |
451 | (perhaps for 2.8.0, perhaps not) will require you to use it as | |
452 | indicated by the current draft. | |
453 | ||
454 | Handling of user-defined type conversion has been overhauled so that | |
455 | type conversion operators are now found and used properly in | |
456 | expressions and function calls. | |
457 | ||
458 | -fno-strict-prototype now only applies to function declarations with | |
459 | "C" linkage. | |
460 | ||
461 | g++ now warns about 'if (x=0)' with -Wparentheses or -Wall. | |
462 | ||
463 | #pragma weak and #pragma pack are supported on System V R4 targets, as | |
464 | are various other target-specific #pragmas supported by gcc. | |
465 | ||
466 | new and delete of const types is now allowed (with no additional | |
467 | semantics). | |
468 | ||
469 | Explicit instantiation of template methods is now supported. Also, | |
470 | 'inline template class foo<int>;' can be used to emit only the vtable | |
471 | for a template class. | |
472 | ||
473 | With -fcheck-new, g++ will check the return value of all calls to | |
474 | operator new, and not attempt to modify a returned null pointer. | |
475 | ||
476 | The template instantiation code now handles more conversions when | |
477 | passing to a parameter that does not depend on template arguments. | |
478 | This means that code like 'string s; cout << s;' now works. | |
479 | ||
480 | Invalid jumps in a switch statement past declarations that require | |
481 | initializations are now caught. | |
482 | ||
483 | Functions declared 'extern inline' now have the same linkage semantics | |
484 | as inline member functions. On supported targets, where previously | |
485 | these functions (and vtables, and template instantiations) would have | |
486 | been defined statically, they will now be defined as weak symbols so | |
487 | that only one out-of-line definition is used. | |
488 | ||
489 | collect2 now demangles linker output, and c++filt has become part of | |
490 | the gcc distribution. | |
491 | \f | |
492 | Noteworthy changes in GCC version 2.6.3: | |
493 | ||
494 | A few more bugs have been fixed. | |
495 | ||
496 | Noteworthy changes in GCC version 2.6.2: | |
497 | ||
498 | A few bugs have been fixed. | |
499 | ||
500 | Names of attributes can now be preceded and followed by double underscores. | |
501 | ||
502 | Noteworthy changes in GCC version 2.6.1: | |
503 | ||
504 | Numerous (mostly minor) bugs have been fixed. | |
505 | ||
506 | The following new configurations are supported: | |
507 | ||
508 | GNU on x86 (instead of treating it like MACH) | |
509 | NetBSD on Sparc and Motorola 68k | |
510 | AIX 4.1 on RS/6000 and PowerPC systems | |
956d6950 | 511 | Sequent DYNIX/ptx 1.x and 2.x. |
fc5a8790 JL |
512 | Both COFF and ELF configurations on AViiON without using /bin/gcc |
513 | Windows/NT on x86 architecture; preliminary | |
514 | AT&T DSP1610 digital signal processor chips | |
515 | i960 systems on bare boards using COFF | |
516 | PDP11; target only and not extensively tested | |
517 | ||
518 | The -pg option is now supported for Alpha under OSF/1 V3.0 or later. | |
519 | ||
520 | Files with an extension of ".c++" are treated as C++ code. | |
521 | ||
522 | The -Xlinker and -Wl arguments are now passed to the linker in the | |
523 | position they were specified on the command line. This makes it | |
524 | possible, for example, to pass flags to the linker about specific | |
525 | object files. | |
526 | ||
527 | The use of positional arguments to the configure script is no longer | |
528 | recommended. Use --target= to specify the target; see the GCC manual. | |
529 | ||
530 | The 386 now supports two new switches: -mreg-alloc=<string> changes | |
531 | the default register allocation order used by the compiler, and | |
532 | -mno-wide-multiply disables the use of the mul/imul instructions that | |
533 | produce 64 bit results in EAX:EDX from 32 bit operands to do long long | |
534 | multiplies and 32-bit division by constants. | |
535 | \f | |
536 | Noteworthy changes in GCC version 2.6.0: | |
537 | ||
538 | Numerous bugs have been fixed, in the C and C++ front-ends, as | |
539 | well as in the common compiler code. | |
540 | ||
541 | This release includes the C, Objective-C, and C++ compilers. However, | |
542 | we have moved the files for the C++ compiler (G++) files to a | |
543 | subdirectory, cp. Subsequent releases of GCC will split these files | |
544 | to a separate TAR file. | |
545 | ||
546 | The G++ team has been tracking the development of the ANSI standard for C++. | |
547 | Here are some new features added from the latest working paper: | |
548 | ||
549 | * built-in boolean type 'bool', with constants 'true' and 'false'. | |
550 | * array new and delete (operator new [] and delete []). | |
551 | * WP-conforming lifetime of temporaries. | |
552 | * explicit instantiation of templates (template class A<int>;), | |
553 | along with an option (-fno-implicit-templates) to disable emission | |
554 | of implicitly instantiated templates, obsoletes -fexternal-templates. | |
555 | * static member constants (static const int foo = 4; within the | |
556 | class declaration). | |
557 | ||
558 | Many error messages have been improved to tell the user more about the | |
559 | problem. Conformance checking with -pedantic-errors has been | |
560 | improved. G++ now compiles Fresco. | |
561 | ||
562 | There is now an experimental implementation of virtual functions using | |
563 | thunks instead of Cfront-style vtables, enabled with -fvtable-thunks. | |
564 | This option also enables a heuristic which causes the compiler to only | |
565 | emit the vtable in the translation unit where its first non-inline | |
566 | virtual function is defined; using this option and | |
567 | -fno-implicit-templates, users should be able to avoid #pragma | |
568 | interface/implementation altogether. | |
569 | ||
570 | Signatures have been added as a GNU C++ extension. Using the option | |
571 | -fhandle-signatures, users are able to turn on recognition of | |
572 | signatures. A short introduction on signatures is in the section | |
573 | `Extension to the C++ Language' in the manual. | |
574 | ||
575 | The `g++' program is now a C program, rather than a shell script. | |
576 | ||
577 | Lots and lots and lots of bugs fixes, in nested types, access control, | |
578 | pointers to member functions, the parser, templates, overload | |
579 | resolution, etc, etc. | |
580 | ||
581 | There have been two major enhancements to the Objective-C compiler: | |
582 | ||
583 | 1) Added portability. It now runs on Alpha, and some problems with | |
584 | message forwarding have been addressed on other platforms. | |
585 | ||
586 | 2) Selectors have been redefined to be pointers to structs like: | |
587 | { void *sel_id, char *sel_types }, where the sel_id is the unique | |
588 | identifier, the selector itself is no longer unique. | |
589 | ||
590 | Programmers should use the new function sel_eq to test selector | |
591 | equivalence. | |
592 | ||
593 | The following major changes have been made to the base compiler and | |
594 | machine-specific files. | |
595 | ||
596 | - The MIL-STD-1750A is a new port, but still preliminary. | |
597 | ||
598 | - The h8/300h is now supported; both the h8/300 and h8/300h ports come | |
599 | with 32 bit IEEE 754 software floating point support. | |
600 | ||
601 | - The 64-bit Sparc (v9) and 64-bit MIPS chips are supported. | |
602 | ||
603 | - NetBSD is supported on m68k, Intel x86, and pc523 systems and FreeBSD | |
604 | on x86. | |
605 | ||
606 | - COFF is supported on x86, m68k, and Sparc systems running LynxOS. | |
607 | ||
608 | - 68K systems from Bull and Concurrent are supported and System V | |
609 | Release 4 is supported on the Atari. | |
610 | ||
611 | - GCC supports GAS on the Motorola 3300 (sysV68) and debugging | |
612 | (assuming GAS) on the Plexus 68K system. (However, GAS does not yet | |
613 | work on those systems). | |
614 | ||
615 | - System V Release 4 is supported on MIPS (Tandem). | |
616 | ||
617 | - For DG/UX, an ELF configuration is now supported, and both the ELF | |
618 | and BCS configurations support ELF and COFF object file formats. | |
619 | ||
620 | - OSF/1 V2.0 is supported on Alpha. | |
621 | ||
622 | - Function profiling is also supported on Alpha. | |
623 | ||
624 | - GAS and GDB is supported for Irix 5 (MIPS). | |
625 | ||
626 | - "common mode" (code that will run on both POWER and PowerPC | |
627 | architectures) is now supported for the RS/6000 family; the | |
628 | compiler knows about more PPC chips. | |
629 | ||
630 | - Both NeXTStep 2.1 and 3 are supported on 68k-based architectures. | |
631 | ||
632 | - On the AMD 29k, the -msoft-float is now supported, as well as | |
633 | -mno-sum-in-toc for RS/6000, -mapp-regs and -mflat for Sparc, and | |
634 | -membedded-pic for MIPS. | |
635 | ||
636 | - GCC can now convert division by integer constants into the equivalent | |
637 | multiplication and shift operations when that is faster than the | |
638 | division. | |
639 | ||
640 | - Two new warning options, -Wbad-function-cast and | |
641 | -Wmissing-declarations have been added. | |
642 | ||
643 | - Configurations may now add machine-specific __attribute__ options on | |
644 | type; many machines support the `section' attribute. | |
645 | ||
646 | - The -ffast-math flag permits some optimization that violate strict | |
647 | IEEE rules, such as converting X * 0.0 to 0.0. | |
648 | \f | |
649 | Noteworthy changes in GCC version 2.5.8: | |
650 | ||
651 | This release only fixes a few serious bugs. These include fixes for a | |
652 | bug that prevented most programs from working on the RS/6000, a bug | |
653 | that caused invalid assembler code for programs with a `switch' | |
654 | statement on the NS32K, a G++ problem that caused undefined names in | |
655 | some configurations, and several less serious problems, some of which | |
656 | can affect most configuration. | |
657 | ||
658 | Noteworthy change in GCC version 2.5.7: | |
659 | ||
660 | This release only fixes a few bugs, one of which was causing bootstrap | |
661 | compare errors on some systems. | |
662 | ||
663 | Noteworthy change in GCC version 2.5.6: | |
664 | ||
665 | A few backend bugs have been fixed, some of which only occur on one | |
666 | machine. | |
667 | ||
668 | The C++ compiler in 2.5.6 includes: | |
669 | ||
670 | * fixes for some common crashes | |
671 | * correct handling of nested types that are referenced as `foo::bar' | |
672 | * spurious warnings about friends being declared static and never | |
673 | defined should no longer appear | |
674 | * enums that are local to a method in a class, or a class that's | |
675 | local to a function, are now handled correctly. For example: | |
676 | class foo { void bar () { enum { x, y } E; x; } }; | |
677 | void bar () { class foo { enum { x, y } E; E baz; }; } | |
678 | ||
679 | Noteworthy change in GCC version 2.5.5: | |
680 | ||
681 | A large number of C++ bugs have been fixed. | |
682 | ||
683 | The fixproto script adds prototypes conditionally on __cplusplus. | |
684 | ||
685 | Noteworthy change in GCC version 2.5.4: | |
686 | ||
687 | A bug fix in passing of structure arguments for the HP-PA architecture | |
688 | makes code compiled with GCC 2.5.4 incompatible with code compiled | |
689 | with earlier versions (if it passes struct arguments of 33 to 64 bits, | |
690 | interspersed with other types of arguments). | |
691 | ||
692 | Noteworthy change in gcc version 2.5.3: | |
693 | ||
694 | The method of "mangling" C++ function names has been changed. So you | |
695 | must recompile all C++ programs completely when you start using GCC | |
696 | 2.5. Also, GCC 2.5 requires libg++ version 2.5. Earlier libg++ | |
697 | versions won't work with GCC 2.5. (This is generally true--GCC | |
698 | version M.N requires libg++ version M.N.) | |
699 | \f | |
700 | Noteworthy GCC changes in version 2.5: | |
701 | ||
702 | * There is now support for the IBM 370 architecture as a target. | |
703 | Currently the only operating system supported is MVS; GCC does not run | |
704 | on MVS, so you must produce .s files using GCC as a cross compiler, | |
705 | then transfer them to MVS to assemble them. This port is not reliable | |
706 | yet. | |
707 | ||
708 | * The Power PC is now supported. | |
709 | ||
710 | * The i860-based Paragon machine is now supported. | |
711 | ||
712 | * The Hitachi 3050 (an HP-PA machine) is now supported. | |
713 | ||
714 | * The variable __GNUC_MINOR__ holds the minor version number of GCC, as | |
715 | an integer. For version 2.5.X, the value is 5. | |
716 | ||
717 | * In C, initializers for static and global variables are now processed | |
718 | an element at a time, so that they don't need a lot of storage. | |
719 | ||
720 | * The C syntax for specifying which structure field comes next in an | |
721 | initializer is now `.FIELDNAME='. The corresponding syntax for | |
722 | array initializers is now `[INDEX]='. For example, | |
723 | ||
724 | char whitespace[256] | |
725 | = { [' '] = 1, ['\t'] = 1, ['\n'] = 1 }; | |
726 | ||
727 | This was changed to accord with the syntax proposed by the Numerical | |
728 | C Extensions Group (NCEG). | |
729 | ||
730 | * Complex numbers are now supported in C. Use the keyword __complex__ | |
731 | to declare complex data types. See the manual for details. | |
732 | ||
733 | * GCC now supports `long double' meaningfully on the Sparc (128-bit | |
734 | floating point) and on the 386 (96-bit floating point). The Sparc | |
735 | support is enabled on on Solaris 2.x because earlier system versions | |
736 | (SunOS 4) have bugs in the emulation. | |
737 | ||
738 | * All targets now have assertions for cpu, machine and system. So you | |
739 | can now use assertions to distinguish among all supported targets. | |
740 | ||
741 | * Nested functions in C may now be inline. Just declare them inline | |
742 | in the usual way. | |
743 | ||
744 | * Packed structure members are now supported fully; it should be possible | |
745 | to access them on any supported target, no matter how little alignment | |
746 | they have. | |
747 | ||
748 | * To declare that a function does not return, you must now write | |
749 | something like this (works only in 2.5): | |
750 | ||
751 | void fatal () __attribute__ ((noreturn)); | |
752 | ||
753 | or like this (works in older versions too): | |
754 | ||
755 | typedef void voidfn (); | |
756 | ||
757 | volatile voidfn fatal; | |
758 | ||
759 | It used to be possible to do so by writing this: | |
760 | ||
761 | volatile void fatal (); | |
762 | ||
763 | but it turns out that ANSI C requires that to mean something | |
764 | else (which is useless). | |
765 | ||
766 | Likewise, to declare that a function is side-effect-free | |
767 | so that calls may be deleted or combined, write | |
768 | something like this (works only in 2.5): | |
769 | ||
770 | int computation () __attribute__ ((const)); | |
771 | ||
772 | or like this (works in older versions too): | |
773 | ||
774 | typedef int intfn (); | |
775 | ||
776 | const intfn computation; | |
777 | ||
778 | * The new option -iwithprefixbefore specifies a directory to add to | |
779 | the search path for include files in the same position where -I would | |
780 | put it, but uses the specified prefix just like -iwithprefix. | |
781 | ||
782 | * Basic block profiling has been enhanced to record the function the | |
783 | basic block comes from, and if the module was compiled for debugging, | |
784 | the line number and filename. A default version of the basic block | |
785 | support module has been added to libgcc2 that appends the basic block | |
786 | information to a text file 'bb.out'. Machine descriptions can now | |
787 | override the basic block support module in the target macro file. | |
788 | ||
789 | New features in g++: | |
790 | ||
791 | * The new flag `-fansi-overloading' for C++. Use a newly implemented | |
792 | scheme of argument matching for C++. It makes g++ more accurately | |
793 | obey the rules set down in Chapter 13 of the Annotated C++ Reference | |
794 | Manual (the ARM). This option will be turned on by default in a | |
795 | future release. | |
796 | ||
797 | * The -finline-debug flag is now gone (it was never really used by the | |
798 | compiler). | |
799 | ||
800 | * Recognizing the syntax for pointers to members, e.g., "foo::*bar", has been | |
801 | dramatically improved. You should not get any syntax errors or incorrect | |
802 | runtime results while using pointers to members correctly; if you do, it's | |
803 | a definite bug. | |
804 | ||
805 | * Forward declaration of an enum is now flagged as an error. | |
806 | ||
807 | * Class-local typedefs are now working properly. | |
808 | ||
809 | * Nested class support has been significantly improved. The compiler | |
810 | will now (in theory) support up to 240 nested classes before hitting | |
811 | other system limits (like memory size). | |
812 | ||
813 | * There is a new C version of the `g++' driver, to replace the old | |
814 | shell script. This should significantly improve the performance of | |
815 | executing g++ on a system where a user's PATH environment variable | |
816 | references many NFS-mounted filesystems. This driver also works | |
817 | under MS-DOS and OS/2. | |
818 | ||
819 | * The ANSI committee working on the C++ standard has adopted a new | |
820 | keyword `mutable'. This will allow you to make a specific member be | |
821 | modifiable in an otherwise const class. | |
822 | \f | |
823 | Noteworthy GCC changes in version 2.4.4: | |
824 | ||
825 | A crash building g++ on various hosts (including m68k) has been | |
826 | fixed. Also the g++ compiler no longer reports incorrect | |
827 | ambiguities in some situations where they do not exist, and | |
828 | const template member functions are now being found properly. | |
829 | ||
830 | Noteworthy GCC changes in version 2.4: | |
831 | ||
832 | * On each target, the default is now to return short structures | |
833 | compatibly with the "usual" compiler on that target. | |
834 | ||
835 | For most targets, this means the default is to return all structures | |
836 | in memory, like long structures, in whatever way is used on that | |
837 | target. Use -freg-struct-return to enable returning short structures | |
838 | (and unions) in registers. | |
839 | ||
840 | This change means that newly compiled binaries are incompatible with | |
841 | binaries compiled with previous versions of GCC. | |
842 | ||
843 | On some targets, GCC is itself the usual compiler. On these targets, | |
844 | the default way to return short structures is still in registers. | |
845 | Use -fpcc-struct-return to tell GCC to return them in memory. | |
846 | ||
847 | * There is now a floating point emulator which can imitate the way all | |
848 | supported target machines do floating point arithmetic. | |
849 | ||
850 | This makes it possible to have cross compilation to and from the VAX, | |
851 | and between machines of different endianness. However, this works | |
852 | only when the target machine description is updated to use the new | |
853 | facilities, and not all have been updated. | |
854 | ||
855 | This also makes possible support for longer floating point types. | |
856 | GCC 2.4 supports extended format on the 68K if you use `long double', | |
857 | for targets that have a 68881. (When we have run time library | |
858 | routines for extended floating point, then `long double' will use | |
859 | extended format on all 68K targets.) | |
860 | ||
861 | We expect to support extended floating point on the i386 and Sparc in | |
862 | future versions. | |
863 | ||
864 | * Building GCC now automatically fixes the system's header files. | |
865 | This should require no attention. | |
866 | ||
867 | * GCC now installs an unsigned data type as size_t when it fixes the | |
868 | header files (on all but a handful of old target machines). | |
869 | Therefore, the bug that size_t failed to be unsigned is fixed. | |
870 | ||
871 | * Building and installation are now completely separate. | |
872 | All new files are constructed during the build process; | |
873 | installation just copies them. | |
874 | ||
875 | * New targets supported: Clipper, Hitachi SH, Hitachi 8300, and Sparc | |
876 | Lite. | |
877 | ||
878 | * A totally new and much better Objective C run time system is included. | |
879 | ||
880 | * Objective C supports many new features. Alas, I can't describe them | |
881 | since I don't use that language; however, they are the same ones | |
882 | supported in recent versions of the NeXT operating system. | |
883 | ||
884 | * The builtin functions __builtin_apply_args, __builtin_apply and | |
885 | __builtin_return let you record the arguments and returned | |
886 | value of a function without knowing their number or type. | |
887 | ||
888 | * The builtin string variables __FUNCTION__ and __PRETTY_FUNCTION__ | |
889 | give the name of the function in the source, and a pretty-printed | |
890 | version of the name. The two are the same in C, but differ in C++. | |
891 | ||
892 | * Casts to union types do not yield lvalues. | |
893 | ||
894 | * ## before an empty rest argument discards the preceding sequence | |
895 | of non-whitespace characters from the macro definition. | |
896 | (This feature is subject to change.) | |
897 | ||
898 | ||
899 | New features specific to C++: | |
900 | ||
901 | * The manual contains a new section ``Common Misunderstandings with | |
902 | GNU C++'' that C++ users should read. | |
903 | ||
904 | * #pragma interface and #pragma implementation let you use the same | |
905 | C++ source file for both interface and implementation. | |
906 | However, this mechanism is still in transition. | |
907 | ||
908 | * Named returned values let you avoid an extra constructor call | |
909 | when a function result has a class type. | |
910 | ||
911 | * The C++ operators <? and >? yield min and max, respectively. | |
912 | ||
913 | * C++ gotos can exit a block safely even if the block has | |
914 | aggregates that require destructors. | |
915 | ||
916 | * gcc defines the macro __GNUG__ when compiling C++ programs. | |
917 | ||
918 | * GNU C++ now correctly distinguishes between the prefix and postfix | |
919 | forms of overloaded operator ++ and --. To avoid breaking old | |
920 | code, if a class defines only the prefix form, the compiler | |
921 | accepts either ++obj or obj++, unless -pedantic is used. | |
922 | ||
923 | * If you are using version 2.3 of libg++, you need to rebuild it with | |
924 | `make CC=gcc' to avoid mismatches in the definition of `size_t'. | |
925 | ||
926 | Newly documented compiler options: | |
927 | ||
928 | -fnostartfiles | |
929 | Omit the standard system startup files when linking. | |
930 | ||
931 | -fvolatile-global | |
932 | Consider memory references to extern and global data items to | |
933 | be volatile. | |
934 | ||
935 | -idirafter DIR | |
936 | Add DIR to the second include path. | |
937 | ||
938 | -iprefix PREFIX | |
939 | Specify PREFIX for later -iwithprefix options. | |
940 | ||
941 | -iwithprefix DIR | |
942 | Add PREFIX/DIR to the second include path. | |
943 | ||
944 | -mv8 | |
945 | Emit Sparc v8 code (with integer multiply and divide). | |
946 | -msparclite | |
947 | Emit Sparclite code (roughly v7.5). | |
948 | ||
949 | -print-libgcc-file-name | |
950 | Search for the libgcc.a file, print its absolute file name, and exit. | |
951 | ||
952 | -Woverloaded-virtual | |
953 | Warn when a derived class function declaration may be an error | |
954 | in defining a C++ virtual function. | |
955 | ||
956 | -Wtemplate-debugging | |
957 | When using templates in a C++ program, warn if debugging is | |
958 | not yet fully available. | |
959 | ||
960 | +eN | |
961 | Control how C++ virtual function definitions are used | |
962 | (like cfront 1.x). | |
963 |