/usr/share/acl2-6.3/acl2-init.lisp is in acl2-source 6.3-5.
This file is owned by root:root, with mode 0o644.
The actual contents of the file can be viewed below.
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870 871 872 873 874 875 876 877 878 879 880 881 882 883 884 885 886 887 888 889 890 891 892 893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956 957 958 959 960 961 962 963 964 965 966 967 968 969 970 971 972 973 974 975 976 977 978 979 980 981 982 983 984 985 986 987 988 989 990 991 992 993 994 995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 1062 1063 1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134 1135 1136 1137 1138 1139 1140 1141 1142 1143 1144 1145 1146 1147 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158 1159 1160 1161 1162 1163 1164 1165 1166 1167 1168 1169 1170 1171 1172 1173 1174 1175 1176 1177 1178 1179 1180 1181 1182 1183 1184 1185 1186 1187 1188 1189 1190 1191 1192 1193 1194 1195 1196 1197 1198 1199 1200 1201 1202 1203 1204 1205 1206 1207 1208 1209 1210 1211 1212 1213 1214 1215 1216 1217 1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228 1229 1230 1231 1232 1233 1234 1235 1236 1237 1238 1239 1240 1241 1242 1243 1244 1245 1246 1247 1248 1249 1250 1251 1252 1253 1254 1255 1256 1257 1258 1259 1260 1261 1262 1263 1264 1265 1266 1267 1268 1269 1270 1271 1272 1273 1274 1275 1276 1277 1278 1279 1280 1281 1282 1283 1284 1285 1286 1287 1288 1289 1290 1291 1292 1293 1294 1295 1296 1297 1298 1299 1300 1301 1302 1303 1304 1305 1306 1307 1308 1309 1310 1311 1312 1313 1314 1315 1316 1317 1318 1319 1320 1321 1322 1323 1324 1325 1326 1327 1328 1329 1330 1331 1332 1333 1334 1335 1336 1337 1338 1339 1340 1341 1342 1343 1344 1345 1346 1347 1348 1349 1350 1351 1352 1353 1354 1355 1356 1357 1358 1359 1360 1361 1362 1363 1364 1365 1366 1367 1368 1369 1370 1371 1372 1373 1374 1375 1376 1377 1378 1379 1380 1381 1382 1383 1384 1385 1386 1387 1388 1389 1390 1391 1392 1393 1394 1395 1396 1397 1398 1399 1400 1401 1402 1403 1404 1405 1406 1407 1408 1409 1410 1411 1412 1413 1414 1415 1416 1417 1418 1419 1420 1421 1422 1423 1424 1425 1426 1427 1428 1429 1430 1431 1432 1433 1434 1435 1436 1437 1438 1439 1440 1441 1442 1443 1444 1445 1446 1447 1448 1449 1450 1451 1452 1453 1454 1455 1456 1457 1458 1459 1460 1461 1462 1463 1464 1465 1466 1467 1468 1469 1470 1471 1472 1473 1474 1475 1476 1477 1478 1479 1480 1481 1482 1483 1484 1485 1486 1487 1488 1489 1490 1491 1492 1493 1494 1495 1496 1497 1498 1499 1500 1501 1502 1503 1504 1505 1506 1507 1508 1509 1510 1511 1512 1513 1514 1515 1516 1517 1518 1519 1520 1521 1522 1523 1524 1525 1526 1527 1528 1529 1530 1531 1532 1533 1534 1535 1536 1537 1538 1539 1540 1541 1542 1543 1544 1545 1546 1547 1548 1549 1550 1551 1552 1553 1554 1555 1556 1557 1558 1559 1560 1561 1562 1563 1564 1565 1566 1567 1568 1569 1570 1571 1572 1573 1574 1575 1576 1577 1578 1579 1580 1581 1582 1583 1584 1585 1586 1587 1588 1589 1590 1591 1592 1593 1594 1595 1596 1597 1598 1599 1600 1601 1602 1603 1604 1605 1606 1607 1608 1609 1610 1611 1612 1613 1614 1615 1616 1617 1618 1619 1620 1621 1622 1623 1624 1625 1626 1627 1628 1629 1630 1631 1632 1633 1634 1635 1636 1637 1638 1639 1640 1641 1642 1643 1644 1645 1646 1647 1648 1649 1650 1651 1652 1653 1654 1655 1656 1657 1658 1659 1660 1661 1662 1663 1664 1665 1666 1667 1668 1669 1670 1671 1672 1673 1674 1675 1676 1677 1678 1679 1680 1681 1682 1683 1684 1685 1686 1687 1688 1689 1690 1691 1692 1693 1694 1695 1696 1697 1698 1699 1700 1701 1702 1703 1704 1705 1706 1707 1708 1709 1710 1711 1712 1713 1714 1715 1716 1717 1718 1719 1720 1721 1722 1723 1724 1725 1726 1727 1728 1729 1730 1731 1732 1733 1734 1735 1736 1737 1738 1739 1740 1741 1742 1743 1744 1745 1746 1747 1748 1749 1750 1751 1752 1753 1754 1755 1756 1757 1758 1759 1760 1761 1762 1763 1764 1765 1766 1767 1768 1769 1770 1771 1772 1773 1774 1775 1776 1777 1778 1779 1780 1781 1782 1783 1784 1785 1786 1787 1788 1789 1790 1791 1792 1793 1794 1795 1796 1797 1798 1799 1800 1801 1802 1803 1804 1805 1806 1807 1808 1809 1810 1811 1812 1813 1814 1815 1816 1817 1818 1819 1820 1821 1822 1823 1824 1825 1826 1827 1828 1829 1830 1831 1832 1833 1834 1835 1836 1837 1838 1839 1840 1841 1842 1843 1844 1845 1846 1847 1848 1849 1850 1851 1852 1853 1854 1855 1856 1857 1858 1859 1860 1861 1862 1863 1864 1865 1866 1867 1868 1869 1870 1871 1872 1873 1874 1875 1876 1877 1878 1879 1880 1881 1882 1883 1884 1885 1886 1887 1888 1889 1890 1891 1892 1893 1894 1895 1896 1897 1898 1899 1900 1901 1902 1903 1904 1905 1906 1907 1908 1909 1910 1911 1912 1913 1914 1915 1916 1917 1918 1919 1920 1921 1922 1923 1924 1925 1926 1927 1928 1929 1930 1931 1932 1933 1934 1935 1936 1937 1938 1939 1940 1941 1942 1943 1944 1945 1946 1947 1948 1949 1950 1951 1952 1953 1954 1955 1956 1957 1958 1959 1960 1961 1962 1963 1964 1965 1966 1967 1968 1969 1970 1971 1972 1973 1974 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 | ; ACL2 Version 6.3 -- A Computational Logic for Applicative Common Lisp
; Copyright (C) 2013, Regents of the University of Texas
; This version of ACL2 is a descendent of ACL2 Version 1.9, Copyright
; (C) 1997 Computational Logic, Inc. See the documentation topic NOTE-2-0.
; This program is free software; you can redistribute it and/or modify
; it under the terms of the LICENSE file distributed with ACL2.
; This program is distributed in the hope that it will be useful,
; but WITHOUT ANY WARRANTY; without even the implied warranty of
; MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
; LICENSE for more details.
; Written by: Matt Kaufmann and J Strother Moore
; email: Kaufmann@cs.utexas.edu and Moore@cs.utexas.edu
; Department of Computer Science
; University of Texas at Austin
; Austin, TX 78701 U.S.A.
; This file cannot be compiled because it changes packages in the middle.
; Allow taking advantage of threads in SBCL, CCL, and Lispworks (where we may
; want to build a parallel version, which needs this to take place). At the
; time that we add (perhaps once again) support for HONS in other lisps besides
; CCL (August, 2012), there has been code developed that depends on mv-let
; being the same as multiple-value-bind; see for example community book
; books/centaur/aig/bddify.lisp, in particular the raw Lisp definition of
; count-branches-to, specifically the use of b* in the labels definition of
; lookup. So we also use multiple-value-bind for mv-let when building the HONS
; version.
#+(or (and sbcl sb-thread) ccl lispworks hons)
(push :acl2-mv-as-values *features*)
; Essay on Parallelism, Parallelism Warts, Parallelism Blemishes, Parallelism
; No-fixes, Parallelism Hazards, and #+ACL2-PAR notes.
; These sources incorporate code for an experimental extension for parallelism
; contributed by David Rager during his master's and Ph.D. dissertation work.
; That extension may be built by setting the :acl2-par feature, for example
; using make (see :DOC compiling-acl2p). The incorporation of code supporting
; parallelism has been carried out while taking great pains to preserve the
; functionality of the ACL2 system proper (i.e., without the experimental
; extension for parallelism).
; At the lowest level, parallel computation is carried out by Lisp threads,
; which provide Lisp-level abstractions of OS threads. A thread can be running
; or blocked (e.g., waiting for a semaphore signal). Threads can create other
; threads. We manage the creation and use of threads to reflect the resources
; we have available, for example the number of available cpu cores according to
; our own tracking.
; The implementation of the multi-threading primitives -- futures, spec-mv-let,
; plet, pargs, pand, and por -- has a dependency structure shown as follows.
; For example, multi-threading primitives are at the base of everything,
; futures and plet/pargs/pand/por are built on top of these primitives, and so
; forth, as indicated by the indentations.
; multi-threading primitives (semaphores, locks, condition variables, etc)
; futures
; spec-mv-let
; waterfall parallelism
; user-level program parallelism (no examples as of April 2012)
; plet, pargs, pand, por
; user-level program parallelism (e.g., community book
; books/parallel/fibonacci.lisp)
; This dependency structure roughly correlates to the following file structure.
; #+acl2-par
; multi-threading-raw.lisp (defines multi-threading primitives)
; parallel-raw.lisp (provides raw Lisp defs. of plet, pargs, pand, and por)
; futures-raw.lisp (defines futures and uses some helper functions
; from parallel-raw.lisp)
; parallel.lisp
; #-acl2-par
; parallel.lisp
; One might wonder how we use threads to execute pieces of parallelism work
; (where "pieces of parallelism work" can mean either futures, bindings of
; plet, arguments of calls surrounded by pargs, or arguments given to pand or
; por). For futures, the story is as follows.
; (1) A primitive adds a piece of parallelism work to one of the two
; parallelism queues (either *future-array* or *work-queue*, as
; appropriate).
;
; (2) The primitive checks to see if there are enough threads already in
; existence to process that piece of parallelism work. If so, the
; primitive returns and execution continues until the result of the
; parallelized computation is needed. If not, the primitive creates one to
; many "worker threads" to consume pieces of parallelism work.
;
; (3) The primitive may eventually need the result from the piece of
; parallelism work. In this case, it will read the value from the piece of
; parallelism work (once it is available) and use it as appropriate. In
; the case that the primitive does not need the value (as can happen with a
; pand/por or a spec-mv-let, when the speculative computation is determined
; to be useless), the primitive will abort (or early terminate) the
; parallel execution of the piece of parallelism work.
;
; When a worker thread is created, it performs the following sequence of
; steps.
;
; (A) Waits until there is a piece of parallelism work to consume. A worker
; thread will wait between 10 and 120 seconds before "giving up", unwinding
; itself, and freeing itself as a resource for the operating system to
; collect. The reader might be tempted to assume that there would
; immediately be work, but this is not guaranteed to be the case (because,
; for efficiency reasons, we typically create a handful more threads than
; are needed).
;
; (B) Waits until there is an idle CPU core available, as determined by our
; resource management using the multi-threading primitives available to us
; in the Lisp (as opposed to trying to tell the operating system how to
; schedule our threads). There is no timeout associated with this wait.
;
; (C) Making it to (C) requires that the thread first made it through (A) and
; then also made it through (B). Thus, the thread has both a piece of
; parallelism work and a CPU core. At this point, the thread executes that
; piece of parallelism work.
;
; (D) Perhaps the piece of parallelism work itself encounters a parallelism
; primitive and decides to further parallelize execution. In this case,
; the worker thread will do (1), (2), and (3), as explained above.
;
; (E) At this point, the worker thread has finished executing the piece of
; parallelism work and stores the result of that execution in the
; appropriate place (e.g., for futures, it stores the execution result in
; the "value" slot of the future).
;
; (F) After performing some cleanup, the thread goes back to (A).
; We use the phrase "Parallelism wart:" to label comments about known issues
; for the #+acl2-par build of ACL2 that we would like to fix, time permitting.
; We also use the phrase "Parallelism blemish:" to identify known issues for
; the #+acl2-par build of ACL2 that we intend to fix only if led to do so by
; user complaints. Finally, we use the phrase "Parallelism no-fix:" to label
; comments about known issues for the #+acl2-par build of ACL2 that we do not
; intend to fix, though we could reclassify these if there are sufficiently
; strong user complaints. Searching through the parallism warts, blemishes,
; and no-fixes could be useful when a user reports a bug in #+acl2-par that
; cannot be replicated in #-acl2-par.
; Parallelism hazards are unrelated to parallelism warts, blemishes, and
; no-fixes. Parallelism hazards are macros or functions that are known to be
; theoretically unsafe when performing multi-threaded execution. We originally
; did not expect users to encounter parallelism hazards (because we should have
; programmed such that the hazards never occur). However, in practice, these
; parallelism hazards are somewhat common and we have disabled the automatic
; warning that occurs everytime a hazard occurs. Once we re-enable that
; warning, in the event that users encounter a parallelism hazard, they will be
; asked to report the associated warning to the ACL2 maintainers. For example,
; if state-global-let* is called while executing concurrently, we want to know
; about it and develop a work-around. See *possible-parallelism-hazards* and
; warn-about-parallelism-hazard for more information.
; #+ACL2-PAR notes contain documentation that only applies to #+acl2-par.
; In an effort to avoid code duplication, we created a definition scheme that
; supports defining both serial and parallel versions of a function with one
; call to a defun-like macro. See the definitions of @par-mappings and
; defun@par for an explanation of this scheme.
; Developer note on emacs and parallelism. When comparing with versions up
; through svn revision 335 (May 27, 2011), it may be useful to ignore "@par"
; when ignoring whitespace with meta-x compare-windows in emacs.
; (setq compare-windows-whitespace "\\(@par\\)?[ \t\n]+")
; To revert to the default behavior:
; (setq compare-windows-whitespace
; "\\(\\s-\\|
; \\)+")
; We indent certain calls as we indent calls of defun. (These forms are in
; emacs/emacs-acl2.el.)
; (put 'er@par 'lisp-indent-function 'defun)
; (put 'warning$@par 'lisp-indent-function 'defun)
; To revert:
; (put 'er@par 'lisp-indent-function nil)
; (put 'warning$@par 'lisp-indent-function nil)
; Only allow the feature :acl2-par in environments that support
; multi-threading. Keep this in sync with the error message about CCL,
; Lispworks, and SBCL in set-parallel-execution-fn and with :doc
; compiling-acl2p. If we add support for non-ANSI GCL, consider providing a
; call of reset-parallelism-variables analogous to the one generated by setting
; *reset-parallelism-variables* in our-abort.
#+(and acl2-par (not ccl) (not (and sbcl sb-thread)) (not lispworks))
(error "It is currently illegal to build the parallel version of ACL2 in this
Common Lisp. See source file acl2-init.lisp for this error message,
which is preceded by information (Lisp features) indicating the legal Lisp
implementations.")
#+akcl
(setq si:*notify-gbc* t)
; The following has been superseded; see the section on reading characters from
; files in acl2.lisp.
; ; Dave Greve reported a problem: the saved_acl2 script in CLISP had characters
; ; that, contrary to expectation, were not being interpreter as newlines. The
; ; CLISP folks explained that "CUSTOM:*DEFAULT-FILE-ENCODING* defaults to :DOS
; ; on cygwin, so #\Newline is printed as '\r\n' (CRLF)." We expect that the
; ; following setting will fix the problem; Dave tried an experiment for us that
; ; seemed to validate this expectation.
; #+clisp
; (setq CUSTOM:*DEFAULT-FILE-ENCODING* :unix)
#+(and lispworks (not acl2-par))
(setq system::*stack-overflow-behaviour*
; The following could reasonably be nil or :warn. David Rager did some
; experiments suggesting that ACL2(p) regressions (as of July 2011) may be more
; robust with the :warn setting. However, that setting causes warnings during
; the build, and it's easy to imagine that ACL2 users would also see that
; cryptic warning -- very un-ACL2-like! So we use nil rather than :warn here.
nil)
#+(and lispworks acl2-par)
(setq system:*stack-overflow-behaviour*
; Since a setting of nil is at least sometimes (if not always) ignored when
; safety is set to 0 (according to an email communication between David Rager
; and Martin Simmons), we choose to use the warn setting for the #+acl2-par
; build.
:warn)
; We have observed a significant speedup with Allegro CL when turning off
; its cross-referencing capability. Here are the times before and after
; evaluating the setq form below, in an example from Dave Greve that spends
; a lot of time loading compiled files.
;
; 165.43 seconds realtime, 163.84 seconds runtime.
; 120.23 seconds realtime, 118.32 seconds runtime.
;
; The user is welcome to edit the form below. Note that it doesn't seem to
; affect the profiler.
#+allegro
(setq excl::*record-xref-info* nil
excl::*load-xref-info* nil
excl::*record-source-file-info* nil
excl::*load-source-file-info* nil)
; Create the packages we use.
(load "acl2.lisp")
; We allow ACL2(h) code to take advantage of Ansi CL features. It's
; conceivable that we don't need this restriction (which only applies to GCL),
; but it doesn't currently seem worth the trouble to figure that out.
#+(and hons (not cltl2))
(progn
(format t "~%ERROR: It is illegal to build the experimental HONS version
of ACL2 in this non-ANSI Common Lisp.~%~%")
(acl2::exit-lisp))
; Fix a bug in SBCL 1.0.49 (https://bugs.launchpad.net/bugs/795705), thanks to
; patch provided by Nikodemus Siivola.
#+sbcl
(in-package :sb-c)
#+sbcl
(when (equal (lisp-implementation-version) "1.0.49")
(without-package-locks
(defun undefine-fun-name (name)
(when name
(macrolet ((frob (type &optional val)
`(unless (eq (info :function ,type name) ,val)
(setf (info :function ,type name) ,val))))
(frob :info)
(frob :type (specifier-type 'function))
(frob :where-from :assumed)
(frob :inlinep)
(frob :kind)
(frob :macro-function)
(frob :inline-expansion-designator)
(frob :source-transform)
(frob :structure-accessor)
(frob :assumed-type)))
(values))
))
; Fix a bug in CMUCL 20D. It seems sad to test (reverse "") twice, but
; attempts to avoid that produced warnings about variable *our-old-reverse*
; being undefined, even when using with-compilation-unit.
#+cmucl
(progn
(when (null (ignore-errors (reverse "")))
(defconstant *our-old-reverse* (symbol-function 'reverse)))
(without-package-locks
(when (boundp '*our-old-reverse*)
(defun reverse (x)
(if (equal x "")
""
(funcall (symbol-value '*our-old-reverse*) x)))
(compile 'reverse))))
; WARNING: The next form should be an in-package (see in-package form for sbcl
; just above).
; Now over to the "ACL2" package for the rest of this file.
(in-package "ACL2")
(defconstant *current-acl2-world-key*
; *Current-acl2-world-key* is the property used for the current-acl2-world
; world. We formerly used a defvar, but there seemed to be no reason to use a
; special variable, which Gary Byers has pointed out takes about 5 instructions
; to read in CCL in order to check if a thread-local binding is dynamically in
; effect. So we tried using a defconstant. Unfortunately, that trick failed
; in Allegro CL; even if we use a boundp test to guard the defconstant, when we
; used make-symbol to create the value; the build failed. So the value is now
; an interned symbol.
'acl2_invisible::*CURRENT-ACL2-WORLD-KEY*)
#+cltl2
(when (not (boundp 'COMMON-LISP::*PRINT-PPRINT-DISPATCH*))
; Many improvements were made to ANSI GCL in May, 2013. If
; COMMON-LISP::*PRINT-PPRINT-DISPATCH* is unbound, then something is wrong with
; this Lisp. In particular, with-standard-io-syntax might not work correctly.
(format t
"ERROR: We do not support building ACL2 in~%~
a host ANSI Common Lisp when variable ~s is unbound. Please~%~
obtain a more recent version of your Lisp implementation."
'COMMON-LISP::*PRINT-PPRINT-DISPATCH*)
(exit-lisp))
#+(and gcl cltl2)
; Deal with undefined cltl2 symbols in ANSI GCL, using values that would be
; assigned by with-standard-io-syntax.
(loop for pair in '((COMMON-LISP::*PRINT-LINES* . nil)
(COMMON-LISP::*PRINT-MISER-WIDTH* . nil)
(COMMON-LISP::*PRINT-RIGHT-MARGIN* . nil)
(COMMON-LISP::*READ-EVAL* . t))
when (not (boundp (car pair)))
do (progn (proclaim `(special ,(car pair)))
(setf (symbol-value (car pair))
(cdr pair))))
; It is a mystery why the following proclamation is necessary, but it
; SEEMS to be necessary in order to permit the interaction of tracing
; with the redefinition of si::break-level.
#+akcl
(declaim (special si::arglist))
#+akcl
(let ((v (symbol-function 'si::break-level)))
(setf (symbol-function 'si::break-level)
(function
(lambda (&rest rst)
(format t "~%Raw Lisp Break.~%")
(apply v rst)))))
(defun system-call (string arguments)
; Warning: Keep this in sync with system-call+.
#+gcl
(si::system
(let ((result string))
(dolist
(x arguments)
(setq result (concatenate 'string result " " x)))
result))
#+lispworks
(system::call-system
(let ((result string))
(dolist
(x arguments)
(setq result (concatenate 'string result " " x)))
result))
#+allegro
(let ((result string))
(dolist
(x arguments)
(setq result (concatenate 'string result " " x)))
#-unix
(excl::shell result)
#+unix
; In Allegro CL in Unix, we can avoid spawning a new shell by calling run-shell-command
; on a simple vector. So we parse the resulting string "cmd arg1 ... argk" and
; run with the simple vector #(cmd cmd arg1 ... argk).
(excl::run-shell-command
(let ((lst nil)
(len (length result))
(n 0))
(loop
(if (>= n len) (return)) ; else get next word
(let ((start n)
(ch (char result n)))
(cond
((member ch '(#\Space #\Tab))
(setq n (1+ n)))
(t (loop
(if (or (>= n len)
(member (setq ch (char result n))
'(#\Space #\Tab)))
(return)
(setq n (1+ n))))
(setq lst (cons (subseq result start n)
lst))))))
(setq result (nreverse lst))
(setq result (coerce (cons (car result) result) 'vector)))))
#+cmu
(ext:process-exit-code
(common-lisp-user::run-program string arguments :output t))
#+sbcl
(sb-ext:process-exit-code
(sb-ext:run-program string arguments :output t :search t))
#+clisp
(let ((result (ext:run-program string :arguments arguments)))
(or result 0))
#+ccl
(let* ((proc (ccl::run-program string arguments :output t))
(status (multiple-value-list (ccl::external-process-status proc))))
(if (not (and (consp status)
(eq (car status) :EXITED)
(consp (cdr status))
(integerp (cadr status))))
1 ; just some non-zero exit code here
(cadr status)))
#-(or gcl lispworks allegro cmu sbcl clisp ccl)
(declare (ignore string arguments))
#-(or gcl lispworks allegro cmu sbcl clisp ccl)
(error "SYSTEM-CALL is not yet defined in this Lisp."))
(defun copy-acl2 (dir)
(system-call
"cp"
(append '("makefile"
"acl2.lisp"
"acl2-check.lisp"
"acl2-fns.lisp"
"init.lisp"
"acl2-init.lisp")
(append (let ((result (list (format nil "~a" dir))))
(dolist
(x *acl2-files*)
(setq result
(cons (format nil "~a.lisp" x)
result)))
result)))))
(defun our-probe-file (filename)
; Use this function instead of probe-file if filename might be a directory.
; We noticed that GCL 2.6.7 on 64-bit Linux doesn't recognize directories with
; probe-file. So we use directory instead, which we have found to work in both
; 32-bit and 64-bit Linux environments.
; BUG: It appears that this function returns nil in GCL 2.6.7 when given an
; existing but empty directory.
#+gcl
(or (probe-file filename)
(let ((x (and (not (equal filename ""))
(cond ((eql (char filename (1- (length filename))) #\/)
filename)
(t (concatenate 'string filename "/"))))))
(directory x)))
#-gcl
(probe-file filename))
(defun copy-distribution (output-file source-directory target-directory
&optional
(all-files "all-files.txt")
(use-existing-target nil))
; We check that all files and directories exist that are supposed to exist. We
; cause an error if not, which ultimately will cause the Unix process that
; calls this function to return an error status, thus halting the make of which
; this operation is a part. Wart: Since probe-file does not check names with
; wildcards, we skip those.
; Note: This function does not actually do any copying or directory creation;
; rather, it creates a file that can be executed.
; FIRST, we make sure we are in the expected directory.
(cond ((not (and (stringp source-directory)
(not (equal source-directory ""))))
(error "The source directory specified for COPY-DISTRIBUTION~%~
must be a non-empty string, but~%~s~%is not."
source-directory)))
(cond ((not (and (stringp target-directory)
(not (equal target-directory ""))))
(error "The target directory specified for COPY-DISTRIBUTION~%must ~
be a non-empty string, but~%~s~%is not. (If you invoked ~
\"make copy-distribution\", perhaps you forgot to set DIR.)"
target-directory)))
(cond ((eql (char source-directory (1- (length source-directory))) #\/)
; In this code we treat all directories as names without the trailing slash.
(setq source-directory
(subseq source-directory 0 (1- (length source-directory))))))
(cond ((not (equal (our-truename (format nil "~a/" source-directory) :safe)
(our-truename
""
"Note: Calling OUR-TRUENAME from COPY-DISTRIBUTION.")))
(error "We expected to be in the directory~%~s~%~
but instead are apparently in the directory~%~s .~%~
Either issue, in Unix, the command~%~
cd ~a~%~
or else edit the file (presumably, makefile) from~%~
which the function COPY-DISTRIBUTION was called,~%~
in order to give it the correct second argument."
source-directory
(our-truename "" t)
source-directory)))
; Next, check that everything exists that is supposed to.
(cond ((and (not use-existing-target)
(our-probe-file target-directory))
(error "Aborting copying of the distribution. The target ~%~
distribution directory~%~s~%~
already exists! You may wish to execute the following~%~
Unix command to remove it and all its contents:~%~
rm -r ~a"
target-directory target-directory)))
(format t "Checking that distribution files are all present.~%")
(let (missing-files)
(with-open-file
(str (concatenate 'string source-directory "/" all-files)
:direction :input)
(let (filename (dir nil))
(loop (setq filename (read-line str nil))
(cond
((null filename) (return))
((or (equal filename "")
(equal (char filename 0) #\#)))
((find #\Tab filename)
(error "Found a line with a Tab in it: ~s" filename))
((find #\Space filename)
(error "Found a line with a Space in it: ~s" filename))
((find #\* filename)
(format t "Skipping wildcard file name, ~s.~%" filename))
((eql (char filename (1- (length filename))) #\:)
; No need to check for directories here; they'll get checked elsewhere. But
; it's harmless enough to do so.
(let* ((new-dir (subseq filename 0 (1- (length filename))))
(absolute-dir
(format nil "~a/~a" source-directory new-dir)))
(cond
((our-probe-file absolute-dir)
(setq dir new-dir))
(t
(setq missing-files
(cons absolute-dir missing-files))
(error "Failed to find directory ~a ."
absolute-dir)))))
(t (let ((absolute-filename
(if dir
(format nil "~a/~a/~a" source-directory dir filename)
(format nil "~a/~a" source-directory filename))))
(cond
((not (our-probe-file absolute-filename))
(setq missing-files
(cons absolute-filename missing-files))
(format t "Failed to find file ~a.~%" absolute-filename)))))))))
(cond
(missing-files
(error "~%Missing the following files (and/or directories):~%~s"
missing-files))
(t (format t "Distribution files are all present.~%"))))
(format t "Preparing to copy distribution files from~%~a/~%to~%~a/ .~%"
source-directory target-directory)
(let (all-dirs)
; In this pass, we look only for directory names.
(with-open-file
(str (concatenate 'string source-directory "/" all-files)
:direction :input)
(let (filename)
(loop (setq filename (read-line str nil))
(cond
((null filename) (return))
((or (equal filename "")
(equal (char filename 0) #\#)))
((find #\Tab filename)
(error "Found a line with a Tab in it: ~s" filename))
((find #\Space filename)
(error "Found a line with a Space in it: ~s" filename))
((eql (char filename (1- (length filename))) #\:)
(setq all-dirs
(cons (subseq filename 0 (1- (length filename)))
all-dirs)))))))
; In the final pass we do our writing.
(with-open-file
(str (concatenate 'string source-directory "/" all-files)
:direction :input)
(with-open-file
(outstr output-file :direction :output)
(let (filename (dir nil))
(if (not use-existing-target)
(format outstr "mkdir ~a~%~%" target-directory))
(loop (setq filename (read-line str nil))
(cond
((null filename) (return))
((or (equal filename "")
(equal (char filename 0) #\#)))
((eql (char filename (1- (length filename))) #\:)
(setq dir (subseq filename 0 (1- (length filename))))
(format outstr "~%mkdir ~a/~a~%"
target-directory dir))
((null dir)
(cond ((not (member filename all-dirs
:test 'equal))
(format outstr "cp -p ~a/~a ~a~%"
source-directory
filename
target-directory))))
(t
(cond ((not (member (format nil "~a/~a"
dir filename)
all-dirs
:test 'equal))
(format outstr "cp -p ~a/~a/~a ~a/~a~%"
source-directory
dir
filename
target-directory
dir)))))))))
(format t "Finished creating a command file for copying distribution files.")))
(defun make-tags ()
#-(or ccl sbcl cmu)
; We disallow ccl and sbcl for the following check. We have found that the
; result of the system-call is a process, (typep <result> 'external-process) in
; ccl and (typep <result> 'sb-impl::process) in sbcl, which can probably be
; made to yield the status. But the status is 0 even for commands not found,
; so why bother? Since cmucl seems to fall victim in the same way as sbcl, we
; treat these two the same here.
(when (not (eql (system-call "which" '("etags")) 0))
(format t "SKIPPING etags: No such program is in the path.")
(return-from make-tags 1))
(system-call "etags"
(let* ((fmt-str
#+(or cmu sbcl clisp ccl) "~a.lisp"
#-(or cmu sbcl clisp ccl) " ~a.lisp")
(lst (append '("acl2.lisp"
"acl2-check.lisp"
"acl2-fns.lisp"
"init.lisp"
"acl2-init.lisp"
"akcl-acl2-trace.lisp"
"allegro-acl2-trace.lisp"
"openmcl-acl2-trace.lisp")
(let ((result nil))
(dolist
(x *acl2-files*)
(setq result
(cons (format nil fmt-str x)
result)))
(reverse result)))))
; We want to be sure to include the *-raw.lisp files even if we are not
; building the hons version, in order to assist in maintaining both versions.
(append lst (list "hons-raw.lisp"
"memoize-raw.lisp"
"multi-threading-raw.lisp"
"futures-raw.lisp"
"parallel-raw.lisp"
"serialize-raw.lisp")))))
(defvar *saved-build-date-lst*)
(defvar *saved-mode*)
(defun svn-revision-from-line (s)
; S is a string such as "$Revision: 1053 $" (as in acl2-startup-info.txt) or
; "Revision: 1998" (as printed by "svn info"). In general, it is a string for
; which we want the object that is read immediately after the first #\: . If
; there is none, then we return nil.
(let ((p (position #\: s)))
(and p
(read-from-string s nil nil :start (1+ p)))))
(defconstant *acl2-svn-revision-string*
(let ((file "acl2-startup-info.txt"))
(cond ((probe-file file)
(let ((val (with-open-file
(str file :direction :input)
(read str))))
(cond ((eq val :release)
nil)
((stringp val)
(let ((n (svn-revision-from-line val)))
(or n (error "Unexpected error in getting svn revision ~
from string:~%~s~%"
val))
(format nil
"
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
+ WARNING: This is NOT an ACL2 release; it is svn revision ~a. +
+ The authors of ACL2 consider svn distributions to be experimental; +
+ they may be incomplete, fragile, and unable to pass our own +
+ regression. Bug reports should include the following line: +
+ ACL2 svn revision ~a; community books svn revision ~a +
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
"
n n "~a")))
(t (error "Illegal value in file ~s: ~s"
file val)))))
(t (error "File ~s appears not to exist." file)))))
(defvar *saved-string*
(concatenate
'string
"~% ~a built ~a.~
~% Copyright (C) 2013, Regents of the University of Texas"
"~% ACL2 comes with ABSOLUTELY NO WARRANTY. This is free software and you~
~% are welcome to redistribute it under certain conditions. For details,~
~% see the LICENSE file distributed with ACL2.~%"
(or *acl2-svn-revision-string*
; If acl2-startup-info.txt begins with the symbol :release, then
; *acl2-svn-revision-string* is a string with ~a expecting to be bound to the
; svn revision of the books. Otherwise we put "~a" here to be bound to "".
"~a")
"~a"
#+hons
"~%~% Experimental modification for HONS, memoization, and applicative hash~
~% tables.~%"
#+acl2-par
"~%~% Experimental modification for parallel evaluation. Please expect at~
~% most limited maintenance for this version~%"
"~% See the documentation topic ~a for recent changes."
"~% Note: We have modified the prompt in some underlying Lisps to further~
~% distinguish it from the ACL2 prompt.~%"))
(defun maybe-load-acl2-init ()
(let* ((home (our-user-homedir-pathname))
(fl (and home
(probe-file (merge-pathnames home "acl2-init.lsp")))))
(when fl (load fl))))
(defun chmod-executable (sysout-name)
(system-call "chmod" (list "+x" sysout-name)))
(defun saved-build-dates (separator)
(let* ((date-lst (reverse *saved-build-date-lst*))
(result (car date-lst))
(sep (concatenate
'string
(string #\Newline)
(if (eq separator :terminal)
(concatenate
'string
(make-string (+ 3 (length *copy-of-acl2-version*))
:initial-element #\Space)
"then ")
separator))))
(dolist (s (cdr date-lst))
(setq result (concatenate 'string result sep s)))
result))
(defmacro our-with-standard-io-syntax (&rest args)
(cons #-cltl2 'progn
#+cltl2 'with-standard-io-syntax
args))
(defun user-args-string (inert-args &optional (separator '"--"))
; This function is used when saving executable scripts, which may specify that
; certain command line arguments are not to be processed by Lisp (other than to
; affect the value of a variable or function such as
; ccl::*unprocessed-command-line-arguments*; see books/oslib/argv.lisp). Also
; see :doc save-exec. A common convention is that arguments after `--' are not
; processed by Lisp.
(cond ((null inert-args)
"\"$@\"")
((eq inert-args t)
(concatenate 'string separator " \"$@\""))
(t
(concatenate 'string separator " " inert-args " \"$@\""))))
(defmacro write-exec-file (stream prefix string &rest args)
; Prefix is generally nil, but can be (string . fmt-args). String is the
; actual command invocation, with the indicated format args, args.
`(our-with-standard-io-syntax ; Thus, we hope that progn is OK for GCL!
(format ,stream "#!/bin/sh~%~%")
(format ,stream
"# Saved ~a~%~%"
(saved-build-dates "# then "))
,@(and prefix
`((format ,stream ,(car prefix) ,@(cdr prefix))))
(format ,stream
; We generally take Noah Friedman's suggestion of using "exec" since there is
; no reason to keep the saved_acl2 shell script in the process table. However,
; we have seen Windows write out "C:..." as the path of the executable, because
; that is what truename produces. But then the "exec" seems to consider that
; to be a relative pathname, and invoking ./saved_acl2 thus fails. So we
; eliminate the "exec" in Windows; we have found that this works fine, at least
; for GCL and SBCL.
#-mswindows
(concatenate 'string "exec " ,string)
#+mswindows
,string
,@args)))
(defun proclaim-files (&optional outfilename infilename infile-optional-p)
; IMPORTANT: This function assumes that the defconst and defmacro forms in the
; given files have already been evaluated. One way to achieve this state of
; affairs, of course, is to load the files first.
(when (and outfilename infilename)
(error "It is illegal to supply non-nil values for both optional ~
arguments of proclaim-files."))
(when (not *do-proclaims*)
(return-from proclaim-files nil))
(cond
(outfilename
(format t
"Writing proclaim forms for ACL2 source files to file ~s.~%"
outfilename))
(t
(when infilename
(cond ((probe-file infilename)
(format t
"Loading nontrivial generated file of proclaim forms, ~
~s...~%"
infilename)
(load infilename)
(format t
"Completed load of ~s.~%"
infilename)
(return-from proclaim-files nil))
(infile-optional-p) ; fall through as though infilename is nil
(t (error "File ~s is to be loaded by proclaim-files, but does ~
not exist."
infilename))))
(format t
"Generating and evaluating proclaim forms for ACL2 source ~
files.~%")))
(let (str)
(when outfilename
(if (probe-file outfilename)
(delete-file outfilename))
(or (setq str (safe-open outfilename :direction :output))
(error "Unable to open file ~s for output." outfilename)))
; It is tempting to print an in-package form, but we leave that task to
; proclaim-file, which presumably finds the first form to be an in-package
; form.
(dolist (fl *acl2-files*)
(proclaim-file (format nil "~a.lisp" fl) str))
(when str ; equivalently, when outfilename is non-nil
(close str))))
(defun insert-string (s)
(cond ((null s) "")
(t (concatenate 'string " " s))))
#+gcl
(defvar *saved-system-banner*
; This variable is only used in GCL 2.6.9 and later, and the following comments
; pertain only to that case.
; Set this variable to nil before save-exec in order to save an image without a
; GCL startup banner, as this will leave si::*system-banner* unbound; see
; below.
; ACL2 keeps this value at nil except when acl2-default-restart unbinds
; si::*system-banner*, in which case *saved-system-banner* is set to the value
; of si::*system-banner* just before that unbinding takes place. When
; save-exec saves an image, it first checks whether si::*system-banner* is
; unbound and *saved-system-banner* is non-nil, in which case it sets
; si::*system-banner* to *saved-system-banner*. Even if si::*system-banner* is
; bound, *saved-system-banner* is set to nil before saving an image.
nil)
#+akcl
(defun save-acl2-in-akcl-aux (sysout-name gcl-exec-name
write-worklispext
set-optimize-maximum-pages
host-lisp-args
inert-args)
(when (and (gcl-version-> 2 6 9 t)
*saved-system-banner*)
(when (not (boundp 'si::*system-banner*)) ; true, unless user intervened
(setq si::*system-banner* *saved-system-banner*))
(setq *saved-system-banner* nil))
(if (and write-worklispext (probe-file "worklispext"))
(delete-file "worklispext"))
(let* ((ext "gcl")
(ext+
; We deal with the apparent fact that Windows implementations of GCL append
; ".exe" to the filename created by save-system.
#+mswindows "gcl.exe"
#-mswindows "gcl")
(gcl-exec-file
(unix-full-pathname gcl-exec-name ext+)))
(if write-worklispext
(with-open-file (str "worklispext" :direction :output)
(format str ext+)))
(if (probe-file sysout-name)
(delete-file sysout-name))
(if (probe-file gcl-exec-file)
(delete-file gcl-exec-file))
(with-open-file (str sysout-name :direction :output)
(write-exec-file str nil "~s~s ~a~%"
gcl-exec-file
(insert-string host-lisp-args)
(user-args-string inert-args)))
(cond ((and set-optimize-maximum-pages
(boundp 'si::*optimize-maximum-pages*))
; We follow a suggestion of Camm Maguire by setting
; 'si::*optimize-maximum-pages* to t just before the save. We avoid the
; combination of 'si::*optimize-maximum-pages* and sgc-on for GCL versions
; through 2.6.3, because of problematic interactions between SGC and
; si::*optimize-maximum-pages*. This issue has been fixed starting with GCL
; 2.6.4. Since si::*optimize-maximum-pages* is only bound starting with
; sub-versions of 2.6, the problem only exists there.
(cond ((or (not (fboundp 'si::sgc-on))
(gcl-version-> 2 6 3))
(setq si::*optimize-maximum-pages* t)))))
(chmod-executable sysout-name)
(si::save-system (concatenate 'string sysout-name "." ext))))
#+akcl
(defun save-acl2-in-akcl (sysout-name gcl-exec-name
&optional mode do-not-save-gcl)
(setq *saved-mode* mode)
(setq *acl2-allocation-alist*
; If *acl2-allocation-alist* is rebound before allocation is done in
; si::*top-level-hook*, e.g., if it is bound in one's init.lisp or
; acl2-init.lsp file, then such binding will override this one. The package
; name shouldn't matter for the keys in user's alist, but in the code below we
; need to keep 'hole in the ACL2 package because we refer to it below.
; Historical Comments:
; Where did these numbers come from? At CLInc we have used the numbers from
; the non-small-p case for some time, and they seem satisfactory. When we
; moved to a "small" image in Version 1.8, we wanted to have roughly the same
; number of free cells as we've had all along, as a default. The cons number
; below is obtained by seeing how many pages we had free (pages in use
; multiplied by percent free, as shown by (room)) the last time we built ACL2,
; before modifying ACL2 to support small images, and adding that to the number
; of pages in use in the small image when no extra pages were allocated at
; start-up. The total was 2917, so that is what we use below. The relocatable
; size is rather arbitrary, and the hole size has been suggested by Bill
; Schelter. Finally, the other numbers were unchanged when we used the same
; algorithm described above for cons (except for fixnum, which came out to 99
; -- close enough!).
; Warning: as of this writing (5/95), there are versions of Linux in which the
; page size is half of that in GCL on a Sparc. In that case, we should double
; the number of pages in each case in order to have the same amount of free
; objects available. We do this below; see the next comment. (We assume that
; there are still the same number of bytes per object; at least, in one
; instance in Linux that appears to be the case for cons, namely, 12 bytes per
; cons.)
; Additional comments during Version_2.9 development:
; When built with GCL 2.6.1-38 and *acl2-allocation-alist* = nil, we have:
; ACL2>(room)
;
; 4972/4972 61.7% CONS RATIO LONG-FLOAT COMPLEX STRUCTURE
; 133/274 14.0% FIXNUM SHORT-FLOAT CHARACTER RANDOM-STATE READTABLE NIL
; 210/462 97.5% SYMBOL STREAM
; 1/2 37.2% PACKAGE
; 69/265 1.0% ARRAY HASH-TABLE VECTOR BIT-VECTOR PATHNAME CCLOSURE FAT-STRING
; 1290/1884 7.4% STRING
; 711/779 0.9% CFUN BIGNUM
; 29/115 82.8% SFUN GFUN CFDATA SPICE NIL
;
; 1302/1400 contiguous (176 blocks)
; 13107 hole
; 5242 0.0% relocatable
;
; 7415 pages for cells
; 27066 total pages
; 93462 pages available
; 10544 pages in heap but not gc'd + pages needed for gc marking
; 131072 maximum pages
;
; ACL2>
; So as an experiment we used some really large numbers below (but not for hole or
; relocatable). They seemed to work well, but see comment just below.
; End of Historical Comments.
(cond
((gcl-version-> 2 6 1)
; In GCL 2.6.5, and in fact starting (we believe) with GCL 2.6.2, GCL does not
; need preallocation to do the job well. We got this opinion after discussions
; with Bob Boyer and Camm Maguire. In a pre-release of Version_2.9, we found
; there was no noticeable change in regression time or image size when avoiding
; preallocation. So it seems reasonable to stop messing with such numbers so
; that they do not become stale and interfere with GCL doing its job.
nil)
(t
`((hole)
(relocatable)
(cons . 10000)
(fixnum . 300)
; Apparently bignums are in CFUN space starting with GCL 2.4.0. So we make
; sure explicitly that there is enough room for bignums. Before GCL 2.4.0,
; bignums are in CONS space so the following should be irrelevant.
(bignum . 800)
(symbol . 500)
(package)
(array . 300)
(string . 2000)
;;(cfun . 32) ; same as bignum
(sfun . 200)))))
; Now adjust if the page size differs from that for GCL/AKCL running on a
; Sparc. See comment above.
(let ((multiplier (/ 4096 si::lisp-pagesize)))
(cond ((not (= multiplier 1))
(setq *acl2-allocation-alist*
(loop for (type . n) in *acl2-allocation-alist*
collect
(cons type
(and n
(round (* multiplier n)))))))))
(setq si::*top-level-hook*
#'(lambda ()
(acl2-default-restart)
(cond
(*acl2-allocation-alist*
; (format
; t
; "Beginning allocations. Set acl2::*acl2-allocation-alist* to NIL~%~
; in ~~/acl2-init.lsp if you must make your running image smaller.~%")
(loop for (type . n) in *acl2-allocation-alist*
when n
do
; (format t "Allocating ~s to ~s.~%" type n)
(let ((x (symbol-name type)))
(cond
((equal x "HOLE")
(si::set-hole-size n))
((equal x "RELOCATABLE")
(si::allocate-relocatable-pages n))
(t (si::allocate type n t)))))))
(lp)))
(load "akcl-acl2-trace.lisp")
; Return to normal allocation growth. Keep this in sync with load-acl2, which
; had presumably already set the allocation growth to be particularly slow.
(loop
for type in
'(cons fixnum symbol array string cfun sfun
; In akcl, at least some versions of it, we cannot call allocate-growth on the
; following two types.
; Camm Maguire has told us on 9/22/2013 that certain allocations for contiguous
; pages, as we now do in acl2.lisp for GCL 2.6.10 and later (which includes GCL
; 2.6.10pre as of 9/22/2013).
; #+gcl contiguous
#+gcl relocatable
)
do
(cond
((or (boundp 'si::*gcl-major-version*) ;GCL 2.0 or greater
(and (boundp 'si::*gcl-version*) ;GCL 1.1
(= si::*gcl-version* 1)))
(si::allocate-growth type 0 0 0 0))
(t (si::allocate-growth type 0 0 0))))
; (print "Start (si::gbc nil)") ;debugging GC
; Camm Maguire suggests leaving the hole size alone for GCL 2.6.10pre as of
; 9/22/2013:
; (si::set-hole-size 500) ; wfs suggestion
; Camm Maguire says (7/04) that "the gc algorithm skips over any pages which
; have not been written to since sgc-on was invoked. So gc really needs to be
; done before turning [sgc] on (not off)...."
(si::gbc t) ; wfs suggestion [at least if we turn on SGC] -- formerly nil
; (don't know why...)
(cond ((fboundp 'si::sgc-on)
(print "Executing (si::sgc-on t)") ;debugging GC
(funcall 'si::sgc-on t)))
; Set the hole to be sufficiently large so that ACL2 can do all the allocations
; quickly when it starts up, without any GC, leaving the desired size hole when
; finished.
(let ((new-hole-size
(or (cdr (assoc 'hole *acl2-allocation-alist*))
(si::get-hole-size))))
(loop for (type . n) in *acl2-allocation-alist*
with space
when (and n
(not (equal (symbol-name type) "HOLE"))
(< (setq space
#+gcl
(cond ;2.0 or later?
((boundp 'si::*gcl-major-version*)
(nth 1 (multiple-value-list
(si::allocated type))))
(t
(caddr (si::allocated type))))
#-gcl
(cond
((equal (symbol-name type)
"RELOCATABLE")
(si::allocated-relocatable-pages))
(t (si::allocated-pages type))))
n))
do (setq new-hole-size (+ new-hole-size (- n space))))
; (print "Set hole size") ;debugging
; Camm Maguire suggests leaving the hole size alone for GCL 2.6.10pre as of
; 9/22/2013:
; (si::set-hole-size new-hole-size)
)
; The calculation above is legacy. Now we increment the hole size to 20% of
; max-pages instead of the default 10%. Camm Maguire says that "Larger values
; allow quick allocation of pages without triggering gc" and that the hole is
; part of the virtual (not resident) memory size, rather than being saved to
; disk.
; Camm Maguire suggests leaving the hole size alone for GCL 2.6.10pre as of
; 9/22/2013:
; (let ((new-size (floor si:*lisp-maxpages* 5)))
; (if (< (si:get-hole-size) new-size)
; (si::set-hole-size new-size)))
; (print (true-listp (w *the-live-state*))) ;swap in the world's pages
; (print "Save the system") ;debugging
(when (not do-not-save-gcl)
(save-acl2-in-akcl-aux sysout-name gcl-exec-name t t nil nil)))
#+akcl
(defun save-exec-raw (sysout-name host-lisp-args inert-args)
(setq *acl2-allocation-alist* nil) ; Don't meddle with allocations.
(setq *acl2-default-restart-complete* nil)
(save-acl2-in-akcl-aux sysout-name sysout-name nil nil host-lisp-args
inert-args))
(defvar *acl2-default-restart-complete* nil)
(defun fix-default-pathname-defaults ()
; Some Lisps save *default-pathname-defaults* and do not reset it at startup.
; According to our experiments:
; - CCL, CMUCL, LispWorks, and GCL retain *default-pathname-defaults*.
; - SBCL and Allegro CL apparently do not retain *default-pathname-defaults*,
; but instead setting it at startup according to the current working
; directory.
; - CLISP sets *default-pathname-defaults* to #P"" at startup.
; But since *default-pathname-defaults* can affect truename, we want it to
; reflect the current working directory.
#+(or ccl cmu gcl lispworks)
(when (pathname-directory *default-pathname-defaults*)
(let ((p (make-pathname)))
(format t "~%Note: Resetting *default-pathname-defaults* to ~s.~%"
p)
(setq *default-pathname-defaults* p)))
nil)
(defvar *print-startup-banner*
; One might want to set this variable to nil in raw Lisp before calling
; save-exec, in order to avoid seeing startup information. We do not comment
; here on whether that is legally appropriate; for example, it suppresses
; copyright information for ACL2 and, for CCL at least, information about the
; host Lisp. We also do not guarantee that this behavior (suppressing printing
; of startup information) is supported for every host Lisp.
; Note that LD always prints some startup information, regardless of the value
; of *print-startup-banner*. To suppress that information, evaluate
; (set-ld-verbose nil state) in the ACL2 loop.
t)
(defvar *lp-ever-entered-p* nil)
(defun acl2-default-restart ()
(if *acl2-default-restart-complete*
(return-from acl2-default-restart nil))
(setq *lp-ever-entered-p* nil)
(#+cltl2
common-lisp-user::acl2-set-character-encoding
#-cltl2
user::acl2-set-character-encoding)
(fix-default-pathname-defaults)
#+ccl
(progn
; In CCL, print greeting now, rather than upon first re-entry to ACL2 loop.
; Here we follow a suggestion from Gary Byers.
(when *print-startup-banner*
(format t "~&Welcome to ~A ~A!~%"
(lisp-implementation-type)
(lisp-implementation-version)))
(setq ccl::*inhibit-greeting* t))
#+gcl
(progn
; Some recent versions of GCL (specifically, 2.6.9 in Sept. 2013) do not print
; the startup banner until we first exit the loop. So we handle that situation
; much as we handle a similar issue for CCL above, following GCL source file
; lsp/gcl_top.lsp.
(when (and *print-startup-banner*
(gcl-version-> 2 6 9 t)
(boundp 'si::*system-banner*))
(format t si::*system-banner*)
(setq *saved-system-banner* si::*system-banner*)
(makunbound 'si::*system-banner*)
(when (boundp 'si::*tmp-dir*)
(format t "Temporary directory for compiler files set to ~a~%"
si::*tmp-dir*))))
#+hons (qfuncall acl2h-init)
(when *print-startup-banner*
(format t
*saved-string*
*copy-of-acl2-version*
(saved-build-dates :terminal)
(if (null *acl2-svn-revision-string*)
""
(qfuncall acl2-books-revision))
(cond (*saved-mode*
(format nil "~% Initialized with ~a." *saved-mode*))
(t ""))
(eval '(latest-release-note-string)) ; avoid possible warning
))
(maybe-load-acl2-init)
(eval `(in-package ,*startup-package-name*))
; The following two lines follow the recommendation in Allegro CL's
; documentation file doc/delivery.htm.
#+allegro (tpl:setq-default *package* (find-package *startup-package-name*))
#+allegro (rplacd (assoc 'tpl::*saved-package*
tpl:*default-lisp-listener-bindings*)
'common-lisp:*package*)
#+allegro (lp)
#+lispworks (lp)
#+ccl (eval '(lp)) ; using eval to avoid compiler warning
(setq *acl2-default-restart-complete* t)
nil)
#+cmu
(defun cmulisp-restart ()
(when *print-startup-banner*
(extensions::print-herald t))
(acl2-default-restart)
(lp))
#+sbcl
(defun sbcl-restart ()
(acl2-default-restart)
; Use eval to avoid style-warning for undefined function LP.
(eval '(lp)))
#+lucid
(defun save-acl2-in-lucid (sysout-name &optional mode)
(setq *saved-mode* mode)
(user::disksave sysout-name :restart-function 'acl2-default-restart
:full-gc t))
#+lispworks
(defun lispworks-save-exec-aux (sysout-name eventual-sysout-name
host-lisp-args inert-args)
; LispWorks support (Dave Fox) pointed out, in the days of LispWorks 4, that we
; need to be sure to call (mp:initialize-multiprocessing) when starting up. Up
; through ACL2 Version_4.2 we did that by making that call in
; acl2-default-restart. But when testing with LispWorks 6.0, we noticed that
; some processes hang, and we wondered if that has to do with the fact that
; (mp:initialize-multiprocessing) does not return. That also got in the way of
; our running (LP) in acl2-default-restart. We experimented with removing
; (mp:initialize-multiprocessing) from acl2-default-restart, instead passing
; :multiprocessing t to system::save-image. But with that change, we noticed
; that upon exiting the ACL2 loop with :q, we got the following rather scary
; message.
; ;; No live processes except internal servers - stopping multiprocessing
; So we have decided not to call :multiprocessing t, and also not to call
; (mp:initialize-multiprocessing) in the #-acl2-par case.
#+acl2-par
(when mp::*multiprocessing*
(send-die-to-worker-threads)
(mp::stop-multiprocessing)
(gc$))
#+acl2-par
(when *lp-ever-entered-p* ; don't print during compliation
(format t
"If you wish to continue using this image, you will need to call ~%~
'mp:initialize-multiprocessing' instead of calling 'lp'. This ~%~
is necessary because of the way multiprocessing works in ~%~
Lispworks.~%~%"))
; We just make a guess that Lispworks can be handled the way that GCL is
; handled.
(if (probe-file "worklispext")
(delete-file "worklispext"))
(let* ((ext "lw")
(ext+
; We deal with the apparent fact that Windows implementations of GCL append
; ".exe" to the filename created by save-system -- and assume, until someone
; tells us otherwise, that Lispworks does similarly.
#+mswindows "lw.exe"
#-mswindows "lw")
(lw-exec-file
(unix-full-pathname sysout-name ext+))
(eventual-lw-exec-file
(unix-full-pathname eventual-sysout-name ext+)))
(with-open-file (str "worklispext" :direction :output)
(format str ext+))
(if (probe-file sysout-name)
(delete-file sysout-name))
(if (probe-file lw-exec-file)
(delete-file lw-exec-file))
(with-open-file (str sysout-name :direction :output)
(write-exec-file str nil
; We pass options "-init -" and "-siteinit -" to inhibit loading init and patch
; files because because we assume that whatever such files were to be loaded,
; were in fact loaded at the time the original Lispworks executable was saved.
; Of course, individual users who doesn't like this decision and know better
; could always edit this script file, i.e., lw-exec-file, in the same spirit as
; changing the underlying Lisp implementation before building ACL2 (again,
; presumably based on knowledge of the host Lisp implementation).
"~s -init - -siteinit -~a ~a~%"
eventual-lw-exec-file
(insert-string host-lisp-args)
(user-args-string inert-args)))
(chmod-executable sysout-name)
(cond ((and system::*init-file-loaded*
system::*complain-about-init-file-loaded*)
; We hope it's fine to save an image when an init-file has been loaded. Maybe
; somebody can explain to us why LispWorks causes a break in such a situation
; by default (which explains the binding of
; system::*complain-about-init-file-loaded* below).
(format t
"Warning: Overriding LispWorks hesitation to save an image~%~
after init-file has been loaded.~%")
(let ((system::*complain-about-init-file-loaded* nil))
(system::save-image lw-exec-file
:restart-function 'acl2-default-restart
#+acl2-par :multiprocessing
#+acl2-par t
:gc t)))
(t (system::save-image lw-exec-file
:restart-function 'acl2-default-restart
#+acl2-par :multiprocessing
#+acl2-par t
:gc t)))))
#+lispworks
(defun save-acl2-in-lispworks (sysout-name mode eventual-sysout-name)
(setq *saved-mode* mode)
(if (probe-file "worklispext")
(delete-file "worklispext"))
(with-open-file (str "worklispext" :direction :output)
(format str "lw"))
(lispworks-save-exec-aux sysout-name eventual-sysout-name
nil nil))
#+lispworks
(defun save-exec-raw (sysout-name host-lisp-args inert-args)
; See the comment above about :multiprocessing t.
(setq *acl2-default-restart-complete* nil)
(lispworks-save-exec-aux sysout-name sysout-name
host-lisp-args inert-args))
#+cmu
(defun save-acl2-in-cmulisp-aux (sysout-name core-name
host-lisp-args inert-args)
(let ((eventual-sysout-core
(unix-full-pathname core-name "core"))
(sysout-core
(unix-full-pathname sysout-name "core")))
(if (probe-file sysout-name)
(delete-file sysout-name))
(if (probe-file eventual-sysout-core)
(delete-file eventual-sysout-core))
(with-open-file ; write to nsaved_acl2
(str sysout-name :direction :output)
(let* ((prog1 (car extensions::*command-line-strings*))
(len (length prog1))
(prog2 (cond ((< len 4)
; If cmucl is installed by extracting to /usr/local/ then the cmucl command is
; simply "lisp" (thanks to Bill Pase for pointing this out).
"lisp")
; The next two cases apply in 18e (and probably earlier) but not 19a (and
; probably later), which has the correct path (doesn't need "lisp" appended).
((equal (subseq prog1 (- len 4) len) "bin/")
(concatenate 'string prog1 "lisp"))
((equal (subseq prog1 (- len 3) len) "bin")
(concatenate 'string prog1 "/lisp"))
(t prog1))))
(write-exec-file str
nil
"~s -core ~s -dynamic-space-size ~s -eval ~
'(acl2::cmulisp-restart)'~a ~a~%"
prog2
eventual-sysout-core
; In our testing for ACL2 Version_6.2 we found that certification failed for
; ACL2(h) built on CMUCL for the book tau/bounders/elementary-bounders.lisp,
; with the error: "CMUCL has run out of dynamic heap space (512 MB)." This
; failure doesn't seem to be fully reproduceable, but it seems safest to
; increase the stack size. Our CMUCL image, even though on 64-bit linux,
; reported the following when we tried a value of 2000 here:
; -dynamic-space-size must be no greater than 1632 MBytes.
; Indeed, we have exceeded that in a version of community book
; books/centaur/gl/solutions.lisp using ACL2(h) built on CMUCL. So we use the
; maximum possible value just below.
1632
(insert-string host-lisp-args)
(user-args-string inert-args))))
(chmod-executable sysout-name)
(system::gc)
(extensions::save-lisp sysout-core :load-init-file nil :site-init nil
; We call print-herald in cmulisp-restart, so that the herald is printed
; before the ACL2-specific information (and before the call of lp).
:print-herald nil)))
#+cmu
(defun save-acl2-in-cmulisp (sysout-name &optional mode core-name)
(setq *saved-mode* mode)
(if (probe-file "worklispext")
(delete-file "worklispext"))
(with-open-file (str "worklispext" :direction :output)
(format str "core"))
(save-acl2-in-cmulisp-aux sysout-name core-name nil nil))
#+cmu
(defun save-exec-raw (sysout-name host-lisp-args inert-args)
(setq *acl2-default-restart-complete* nil)
(save-acl2-in-cmulisp-aux sysout-name sysout-name host-lisp-args inert-args))
#+sbcl
(defvar *sbcl-dynamic-space-size*
; The user is welcome to set this value, either by setting this variable before
; saving an ACL2 image, or by editing the resulting script (e.g., saved_acl2 or
; saved_acl2h). Here we explain the defaults that we provide for this
; variable.
; We observed during development of Version_5.0 that --dynamic-space-size 2000
; is necessary in order to complete an ACL2(h) regression with SBCL 1.0.55 on a
; Mac OS 10.6 laptop; otherwise Lisp dies during a GC when certifying community
; book books/centaur/tutorial/intro.lisp, even with (clear-memoize-tables)
; executed at the start of acl2-compile-file and with (gc$ :full t) executed
; there as well, and also at the start of write-expansion-file and immediately
; before and after include-book-fn in certify-book-fn. We believe that it has
; been necessary to use such a --dynamic-space-size setting even to build ACL2
; (not only ACL2(h)) with SBCL on some platforms, so we decided to use this
; option for ACL2, not just ACL2(h).
; But in December 2012 we found that 2000 is not sufficient using SBCL 1.0.49
; on our 64-bit linux system. Our first such failure was in certifying
; community book
; books/models/y86/y86-two-level-abs/common/x86-state-concrete.lisp in ACL2(h).
; We tried increasing the --dynamic-space-size to 4000, but that was not
; sufficient for community book
; books/models/y86/y86-basic/common/x86-state.lisp; in fact, 8000 also was not
; sufficient for that book. Fortunately, 16000 was sufficient.
; These are unusual books, in that they allocate an array of size 2^32.
; Therefore we only increase the value to 16000 under #+hons; after all, the
; ACL2 regression (as opposed to ACL2(h)) does not certify the y86 books in
; ACL2. If --dynamic-space-size 16000 causes a problem for some ACL2(h) users,
; a simple solution will be for them to edit saved_acl2 or for them to build
; ACL2 after defining this variable to be smaller than 16000 (though some
; community book certifications may fail under under books/models/y86/, which
; are done by default for ACL2(h)).
; On 32-bit systems, 16000 may be too large. We tried it on a 32-bit Linux
; system and got an error upon starting ACL2: "--dynamic-space-size argument is
; out of range: 16000". So we revert to our earlier value of 2000 for such
; systems, even if we are doing an ACL2(h) build. (The y86 books will likely
; fail in this case, but we expect ACL2(h) users will generally be on 64-bit
; systems.)
#+(and x86-64 hons) 16000
#-(and x86-64 hons) 2000)
#+sbcl
(defvar *sbcl-contrib-dir* nil)
#+sbcl
(defun save-acl2-in-sbcl-aux (sysout-name core-name
host-lisp-args
toplevel-args
inert-args)
; Note that host-lisp-args specifies what the SBCL manual calls "runtime
; options", while toplevel-args is what it calls "toplevel options".
(declaim (optimize (sb-ext:inhibit-warnings 3)))
(let ((eventual-sysout-core
(unix-full-pathname core-name "core"))
(sysout-core
(unix-full-pathname sysout-name "core")))
(if (probe-file sysout-name)
(delete-file sysout-name))
(if (probe-file eventual-sysout-core)
(delete-file eventual-sysout-core))
(with-open-file ; write to nsaved_acl2
(str sysout-name :direction :output)
(let* ((prog (car sb-ext:*posix-argv*)))
(write-exec-file
str
; In order to profile, Nikodemus Siivola has told us that we "need to set
; SBCL_HOME to the location of the contribs". Using apropos we found the
; function SB-INT:SBCL-HOMEDIR-PATHNAME that returns the right directory in
; SBCL Version 1.0.23. So we'll use that.
("~a~%"
(let ((contrib-dir
(or
*sbcl-contrib-dir*
(and (boundp 'sb-ext::*core-pathname*)
(ignore-errors
(let* ((core-dir
(pathname-directory
sb-ext::*core-pathname*))
(contrib-dir-pathname
(and (equal (car (last core-dir))
"output")
(make-pathname
:directory
(append (butlast core-dir 1)
(list "contrib"))))))
(and (probe-file contrib-dir-pathname)
(setq *sbcl-contrib-dir*
(namestring contrib-dir-pathname)))))))))
(if contrib-dir
(format nil
"export SBCL_HOME=~s"
contrib-dir)
"")))
; We have observed with SBCL 1.0.49 that "make HTML" fails on our 64-bit linux
; system unless we start sbcl with --control-stack-size 4 [or larger]. The
; default, according to http://www.sbcl.org/manual/, is 2. The problem seems
; to be stack overflow from fmt0, which is not tail recursive. More recently,
; community book centaur/misc/defapply.lisp causes a stack overflow even with
; --control-stack-size 4 (though that might disappear after we added (comp t)
; in a couple of places). Yet more recently, community books
; books/centaur/regression/common.lisp and books/centaur/tutorial/intro.lisp
; fail with --control-stack-size 8, due to calls of def-gl-clause-processor.
; So we use --control-stack-size 16. We might increase 16 to 32 or greater in
; the future.
; See *sbcl-dynamic-space-size* for an explanation of the --dynamic-space-size
; setting below.
; Note that --no-userinit was introduced into SBCL in Version 0.9.13, hence has
; been part of SBCL since 2007 (perhaps earlier). So when Jared Davis pointed
; out this option to us after ACL2 Version_6.2, we started using it in place of
; " --userinit /dev/null", which had not worked on Windows.
"~s --dynamic-space-size ~s --control-stack-size 16 --core ~s~a ~
--end-runtime-options --no-userinit --eval '(acl2::sbcl-restart)'~a ~a~%"
prog
*sbcl-dynamic-space-size*
eventual-sysout-core
(insert-string host-lisp-args)
(insert-string toplevel-args)
(user-args-string inert-args "--end-toplevel-options"))))
(chmod-executable sysout-name)
;; In SBCL 0.9.3 the read-only space is too small for dumping ACL2 on x86,
;; so we have to specify :PURIFY NIL. This will unfortunately result in
;; some core file bloat, and slightly slower startup.
(sb-ext:gc)
(sb-ext:save-lisp-and-die sysout-core
:purify
#+(or x86 x86-64 ppc) nil
#-(or x86 x86-64 ppc) t)))
#+sbcl
(defun save-acl2-in-sbcl (sysout-name &optional mode core-name)
(with-warnings-suppressed
(setq *saved-mode* mode)
(if (probe-file "worklispext")
(delete-file "worklispext"))
(with-open-file (str "worklispext" :direction :output)
(format str "core"))
(save-acl2-in-sbcl-aux sysout-name core-name nil nil nil)))
#+sbcl
(defun save-exec-raw (sysout-name host-lisp-args toplevel-args inert-args)
(with-warnings-suppressed
(setq *acl2-default-restart-complete* nil)
(save-acl2-in-sbcl-aux sysout-name sysout-name host-lisp-args toplevel-args
inert-args)))
#+allegro
(defun save-acl2-in-allegro-aux (sysout-name dxl-name
host-lisp-args inert-args)
(excl:gc t) ; Suggestions are welcome for better gc call(s)!
#+(and allegro-version>= (version>= 4 3))
(progn
(tpl:setq-default *PACKAGE* (find-package "ACL2"))
(setq EXCL:*RESTART-INIT-FUNCTION* 'acl2-default-restart)
#+(and allegro-version>= (version>= 5 0))
(progn
; Allegro 5.0 and later no longer supports standalone images. Instead, one
; creates a .dxl file using dumplisp and then starts up ACL2 using the original
; Lisp executable, but with the .dxl file specified using option -I. Our
; saved_acl2 executable is then a one-line script that makes this Lisp
; invocation. Note that :checkpoint is no longer supported starting in 5.0.
(let* ((eventual-sysout-dxl
(if dxl-name
(unix-full-pathname dxl-name "dxl")
(error "An image file must be specified when building ACL2 in ~
Allegro 5.0 or later.")))
(sysout-dxl
(unix-full-pathname sysout-name "dxl")))
(write-acl2rc (our-pwd))
(with-open-file ; write to nsaved_acl2
(str sysout-name :direction :output)
(write-exec-file
str
nil
; We use ~s instead of ~a below because John Cowles has told us that in Windows
; 98, the string quotes seem necessary for the first string and desirable for
; the second. The string quotes do not hurt on Unix platforms.
; We omit the trailing " -L ~s" for now from the following string, which would
; go with format arg (rc-filename save-dir), because we know of no way in
; Allegro 6.2 to avoid getting Allegro copyright information printed upon :q if
; we start up in the ACL2 read-eval-print loop.
; "~s -I ~s -L ~s ~s~%"
"~s -I ~s~s ~a~%"
(system::command-line-argument 0)
eventual-sysout-dxl
(insert-string host-lisp-args)
(user-args-string inert-args)))
(chmod-executable sysout-name)
(excl:dumplisp :name sysout-dxl)))
#-(and allegro-version>= (version>= 5 0))
(excl:dumplisp :name sysout-name :checkpoint nil))
#-(and allegro-version>= (version>= 4 3))
(progn
(excl:dumplisp :name sysout-name :checkpoint t
:restart-function 'acl2-default-restart)))
#+allegro
(defun save-acl2-in-allegro (sysout-name &optional mode dxl-name)
; Note that dxl-name should, if supplied, be a relative pathname string, not
; absolute.
(setq *saved-mode* mode)
(if (probe-file "worklispext")
(delete-file "worklispext"))
(with-open-file (str "worklispext" :direction :output)
(format str "dxl"))
(load "allegro-acl2-trace.lisp") ; Robert Krug's trace patch
(save-acl2-in-allegro-aux sysout-name dxl-name nil nil))
#+allegro
(defun save-exec-raw (sysout-name host-lisp-args inert-args)
(setq *acl2-default-restart-complete* nil)
(save-acl2-in-allegro-aux sysout-name sysout-name host-lisp-args inert-args))
(defun rc-filename (dir)
(concatenate 'string dir ".acl2rc"))
(defun write-acl2rc (dir)
(let ((rc-filename
(concatenate 'string dir ".acl2rc")))
(if (not (probe-file rc-filename))
(with-open-file ; write to .acl2rc
(str (rc-filename dir) :direction :output)
; We call acl2-default-restart before lp so that the banner will be printed
; and (optionally) ~/acl2-init.lsp file will be loaded before entering the ACL2
; read-eval-print loop.
(format str
"; enter ACL2 read-eval-print loop~%~
(ACL2::ACL2-DEFAULT-RESTART)~%~
#+ALLEGRO (EXCL::PRINT-STARTUP-INFO T)~%~
#+ALLEGRO (SETQ EXCL::*PRINT-STARTUP-MESSAGE* NIL)~%~
(ACL2::LP)~%")))))
#+clisp
(defun save-acl2-in-clisp-aux (sysout-name mem-name host-lisp-args inert-args)
(let ((save-dir (our-pwd))
(eventual-sysout-mem
(unix-full-pathname mem-name "mem"))
(sysout-mem
(unix-full-pathname sysout-name "mem")))
(if (probe-file sysout-name)
(delete-file sysout-name))
(if (probe-file eventual-sysout-mem)
(delete-file eventual-sysout-mem))
(write-acl2rc save-dir)
(with-open-file ; write to nsaved_acl2
(str sysout-name :direction :output)
(write-exec-file str
nil
"~s -i ~s -p ACL2 -M ~s -m ~dMB -E ISO-8859-1~a ~a~%"
(or (ext:getenv "LISP") "clisp")
(rc-filename save-dir)
eventual-sysout-mem
; Here we choose a value of 10 for the -m argument. We have found that without
; setting -m, we get a stack overflow in community book
; books/unicode/read-utf8.lisp at (verify-guards read-utf8-fast ...) when
; running on CLISP 2.34 on Linux. CLISP documentation at
; http://clisp.cons.org/clisp.html#opt-memsize says that it is "common to
; specify 10 MB" for the value of -m; since that suffices to eliminate the
; stack overflow mentioned above, we use that value. Note that we use ~dMB
; instead of ~sMB because the (our-)with-standard-io-syntax wrapper in
; write-exec-file seems to put a decimal point after the number when using ~s,
; and CLISP complains about that when starting up.
10
(insert-string host-lisp-args)
(user-args-string inert-args)))
(chmod-executable sysout-name)
(ext:gc)
(ext:saveinitmem sysout-mem
:quiet nil
; We call acl2-default-restart here, even though above we take pains to call it
; in the .acl2rc file, because someone could edit that file but we still want
; the banner printed.
:init-function 'acl2-default-restart)))
#+clisp
(defun save-acl2-in-clisp (sysout-name &optional mode mem-name)
(setq *saved-mode* mode)
(if (probe-file "worklispext")
(delete-file "worklispext"))
(with-open-file (str "worklispext" :direction :output)
(format str "mem"))
(save-acl2-in-clisp-aux sysout-name mem-name nil nil))
#+clisp
(defun save-exec-raw (sysout-name host-lisp-args inert-args)
(setq *acl2-default-restart-complete* nil)
(save-acl2-in-clisp-aux sysout-name sysout-name host-lisp-args inert-args))
#+ccl
(defun save-acl2-in-ccl-aux (sysout-name core-name
&optional
(host-lisp-args nil save-exec-p)
inert-args)
(let* ((ccl-program0
(or (car ccl::*command-line-argument-list*) ; Gary Byers suggestion
(error "Unable to determine CCL program pathname!")))
(ccl-program (qfuncall pathname-os-to-unix
ccl-program0
(get-os)
*the-live-state*))
(core-name (unix-full-pathname core-name
(pathname-name ccl-program))))
(when (probe-file sysout-name)
; At one point we supplied :if-exists :overwrite in the first argument to
; with-open-file below. Robert Krug reported problems with that solution in
; OpenMCL 0.14 (CCL). A safe solution seems to be simply to delete the file
; before attempting to write to it.
(delete-file sysout-name))
(when (probe-file core-name)
(delete-file core-name))
(with-open-file ; write to nsaved_acl2
(str sysout-name :direction :output)
(write-exec-file str
; Gary Byers has pointed out (Feb. 2009) that:
; In order for REQUIRE (and many other things) to work, the lisp needs
; to know where its installation directory (the "ccl" directory) is.
; (More accurately, the "ccl" logical host has to has its logical-pathname
; translations set up to refer to that directory:)
;
; ? (truename "ccl:")
; #P"/usr/local/src/ccl-dev/"
; So we make an effort to set CCL_DEFAULT_DIRECTORY correctly so that the above
; truename will be correct.
("~a~%"
(let ((default-dir
(or (ccl::getenv "CCL_DEFAULT_DIRECTORY")
(let ((path (our-truename "ccl:")))
(and path
(qfuncall pathname-os-to-unix
(namestring path)
(get-os)
*the-live-state*))))))
(if default-dir
(format nil
"export CCL_DEFAULT_DIRECTORY=~s"
default-dir)
"")))
; See the section on "reading characters from files" in file acl2.lisp for an
; explanation of the -K argument below.
; It is probably important to use -e just below instead of :toplevel-function,
; at least for #+hons. Jared Davis and Sol Swords have told us that it seems
; that with :toplevel-function one gets a new "toplevel" thread at start-up,
; which "plays badly with the thread-local hash tables that make up the hons
; space".
"~s -I ~s~a -K ISO-8859-1 -e ~
\"(acl2::acl2-default-restart)\"~a ~a~%"
ccl-program
core-name
(if save-exec-p
; For an ACL2 built from sources, the saved script will include "-Z 64M"; see
; comment below. But with save-exec, no -Z option will be written. The new
; script can then be expected to invoke ACL2 with the same stack sizes as did
; the original (which had -Z 64M explicitly), unless an explicit -Z option is
; given to save-exec or globals such as
; ccl::*initial-listener-default-control-stack-size* (see community book
; books/centaur/ccl-config.lsp) are set before the save-exec call.
; Turning now to the case of building from sources, as opposed to save-exec:
; We use -Z 64M even though the default for -Z (as of mid-2013) is 2M, in order
; to get larger stacks. We have ample evidence that a larger stack would be
; useful: an ACL2 example from David Russinoff in August 2013 for which 8M was
; insufficient (defining a constant function with body '(...), a quoted list of
; length 65536; our own x86 model requiring 4M for an ACL2 proof using
; def-gl-thm; and more generally, Centaur's routine use of large stacks,
; equivalent to -Z 256M. Not surprisingly, we that performance was not hurt
; using a larger stack size, for two pairs of ACL2(h) regressions as follows.
; We ran one pair of runs on a Linux system with 32GB of RAM, and one pair of
; runs on a MacBook Pro with 8GB of RAM, all in August 2013. For each pair we
; ran with -Z 64M and also omitting -Z (equivalent to using -Z 2M). Our main
; concern was potentially larger backtraces when using (set-debugger-enable
; :bt), as during a regression. We solved that by restricting backtrace counts
; using *ccl-print-call-history-count*.
""
" -Z 64M")
(insert-string host-lisp-args)
(user-args-string inert-args)))
(chmod-executable sysout-name)
(ccl::gc)
(ccl:save-application core-name)))
#+ccl
(defun save-acl2-in-ccl (sysout-name &optional mode core-name)
(setq *saved-mode* mode)
(load "openmcl-acl2-trace.lisp")
(save-acl2-in-ccl-aux sysout-name core-name))
#+ccl
(defun save-exec-raw (sysout-name host-lisp-args inert-args)
(setq *acl2-default-restart-complete* nil)
(save-acl2-in-ccl-aux sysout-name sysout-name host-lisp-args inert-args))
; Since saved-build-date-string is avoided for MCL, we avoid the following too
; (which is not applicable to MCL sessions anyhow).
#-(and mcl (not ccl))
(defun save-acl2 (&optional mode other-info
; Currently do-not-save-gcl is ignored for other than GCL. It was added in
; order to assist in the building of Debian packages for ACL2 based on GCL, in
; case Camm Maguire uses compiler::link.
do-not-save-gcl)
#-akcl (declare (ignore do-not-save-gcl))
#-(or akcl allegro cmu sbcl clisp ccl)
(declare (ignore other-info))
#+akcl
(if (boundp 'si::*optimize-maximum-pages*)
(setq si::*optimize-maximum-pages* nil)) ; Camm Maguire suggestion
; Consider adding something like
; (ccl::save-application "acl2-image" :size (expt 2 24))
; for the Mac.
(load-acl2)
(setq *saved-build-date-lst*
; The call of eval below should avoid a warning in cmucl version 18d. Note
; that saved-build-date-string is defined in interface-raw.lisp.
(list (eval '(saved-build-date-string))))
(eval mode)
(princ "
******************************************************************************
Initialization complete, beginning the check and save.
******************************************************************************
")
(cond
((or (not (probe-file *acl2-status-file*))
(with-open-file (str *acl2-status-file*
:direction :input)
(not (eq (read str nil)
:initialized))))
(error "Initialization has failed. Try INITIALIZE-ACL2 again.")))
#+akcl
(save-acl2-in-akcl "nsaved_acl2" other-info mode do-not-save-gcl)
#+lucid
(save-acl2-in-lucid "nsaved_acl2" mode)
#+lispworks
(save-acl2-in-lispworks "nsaved_acl2" mode other-info)
#+allegro
(save-acl2-in-allegro "nsaved_acl2" mode other-info)
#+cmu
(save-acl2-in-cmulisp "nsaved_acl2" mode other-info)
#+sbcl
(save-acl2-in-sbcl "nsaved_acl2" mode other-info)
#+clisp
(save-acl2-in-clisp "nsaved_acl2" mode other-info)
#+ccl
(save-acl2-in-ccl "nsaved_acl2" mode other-info)
#-(or akcl lucid lispworks allegro clisp ccl cmu sbcl)
(error "We do not know how to save ACL2 in this Common Lisp.")
(format t "Saving of ACL2 is complete.~%"))
(defun generate-acl2-proclaims ()
; See the section "PROCLAIMING" in acl2-fns.lisp.
(let ((filename "acl2-proclaims.lisp"))
(cond (*do-proclaims*
(format t "Beginning load-acl2 and initialize-acl2 on behalf of ~
generate-acl2-proclaims.~%")
(load-acl2 t)
; Use funcall to avoid compiler warning in (at least) CCL.
(qfuncall initialize-acl2 'include-book nil nil t)
(proclaim-files filename))
(t
(if (probe-file filename)
(delete-file filename))
(with-open-file
(str filename :direction :output)
(format str "(in-package \"ACL2\")~%~%")
(format str
"; No proclaims are generated here for this host Lisp.~%"))
nil))))
; The following avoids core being dumped in certain circumstances
; resulting from very hard errors.
#+akcl
(si::catch-fatal 1)
|