Here is the list of configurations that have passed the trip and trap tests. See also ./PROBLEMS.WEB2C. Except as otherwise noted: * REGFIX is either defined or irrelevant * optimization is maximally turned on. Here are email addresses for the `who': elizabeth = elisabet@u.washington.edu kabal = kabal@aldebaran.INSL.McGill.CA karl = karl@cs.umb.edu payne = payne@stsci.edu machine operating system compiler who %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% DEC DS2100 Ultrix 4.2 cc kabal Same diffs in trap test as for the 5000/20. coptimize="-O -Olimit 1000". DEC 5000/20 Ultrix 4.2a cc karl I get these changes from the trap test; it appears to be a compiler bug, as compiling with gcc works ok. People report that they can use MF with no ill effects, though, despite this. 3752c3752 < ### -3062.5o2*4 = -xpart %CAPSULE1237-2500.00002oo+1000o3-32758.99998 --- > ### -3062.5o2*4 = -xpart %CAPSULE1237-2500.00002oo+1000o3+32767.99998 3754c3754 < ### -0.04578o3 = -o4+0.44112oo+8.50041 --- > ### -0.04578o3 = -o4+0.44112oo+11.5 3756c3756 < ### 0.21834o5 = -%CAPSULE1213+0.09631oo+3ooo+1.85596 --- > ### 0.21834o5 = -%CAPSULE1213+0.09631oo+3ooo+2.5109 DEC 5000/20 Ultrix 4.2a gcc 2.1 karl -O fails on tangle (at least; I gave up on -O after that). DEC 5000/240 Ultrix 4.2a rev.47 gcc 2.1 elizabeth Without optimization. HP 9000/[34]00 HP-UX 8.0 cc (builtin or optional) pfs web2cy.c, mf9.c and gftodvi.c must be compiled with +O1, not -O. HP 9000/370 MORE/bsd 4.3 cc karl HP 9000/700 HP-UX 8.05 cc (builtin or optional) pfs +O1 compilation for everything. HP 9000/830 HP-UX 8.05 cc rz32 XCFLAGS="-Aa -D_HPUX_SOURCE -UANSI". mf2.c needed +O1. HP 9000/834 HP-UX 7.0 cc,gcc 1.39 karl IBM RS/6000 AIX xlc karl xlc cannot compile the flex-generated scanner. IBM RT AOS (BSD 4.3) gcc 2.1 karl IBM RT AOS (BSD 4.3) pcc -w karl This doesn't work, since pcc can't handle typedef void *address; extern address xmalloc (); IBM RT AOS (BSD 4.3) hc -w karl -O fails on TeX, at least: trip hangs. Intel 386 Interactive Unix 2.2.1 cc -Xp -D_POSIX_SOURCE karl cc and flex/bison don't cooperate. -O doesn't work on tangle, at least: assembler errors. Intel 386 Interactive Unix 2.2.1 gcc 2.1 karl Sun 3 SunOS 4.1.1 cc karl -O hangs on mf2.c. Sun 3 SunOS 4.0.3 cc, gcc 1.36 elizabeth Sun 3 SunOS 4.1.1 gcc 1.40 karl flex and don't mix. Sun 386i SunOS 4.0.2 cc karl mf4.c breaks the optimizer. Sun 386i SunOS 4.0.2 gcc 1.40 karl flex and don't mix. Sun Sparc SunOS 4.1.1 acc 1.0 1 Mar 91 payne Sun Sparc SunOS 4.1.1 cc karl Sun Sparc SunOS 4.1.1 gcc 1.40 karl flex and don't mix. Below here is list for pre-5.851. andy = andy@csvax.cs.caltech.edu bernd = raichle@azu.informatik.uni-stuttgart.de beebe = beebe@science.utah.edu broman = broman@nosc.mil bells = bells@ibm.com casterln = casterln@are.berkeley.edu corrigan = corrigan@weber.ucsd.edu dick = dick@sprite.stanford.edu d.love = d.love@daresbury.ac.uk dorab = dorab@twinsun.com doug = doug@csd4.csd.uwm.edu dsmith = dsmith@applga.aa.cad.slb.com edh = edh@ux.acs.umn.edu Helmut = root@freeman.phys.nwu.edu hrp = hrp@boring.cray.com karl = karl@cs.umb.edu ken = ken@csis.dit.csiro.au klaus = cnix!klaus@uunet.uu.net jmr = jmr@nada.kth.se kabal = kabal@aldebaran.insl.mcgill.ca lgy = lgy@newton.phys.washington.edu mackay = mackay@cs.washington.edu maj = maj@computer-lab.cambridge.ac.uk mark = mark@comp.vuw.ac.nz mok = mok@kemi.aau.dk morgan = morgan@ics.uci.edu newton = newton@gumby.cs.caltech.edu pfs = p.f.smith@aberdeen.ac.uk piet = piet@cs.ruu.nl spqr = sqpr@ecs.southampton.ac.uk uri = uri@ibm.com wittman = wittman@urz.ku-eichstatt.dbp.de xitikgun = xitikgun@ddathd21.bitnet machine operating system compiler who %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% Acorn R140 RiscIX 1.14 cc (Norcroft C 300/331) maj Apollo DN2500 SR 10.2 cc Helmut Erschbaumer Apollo DN3500 SR 10.2 cc 6.7(316) broman AT&T Unix PC gcc andy Define ANSI. The big TeX is too big. Convex C1-XP 4.2bsd/8.0 vc 3.0 doug cc -O also works, except web2c itself fails. Compiled with vc -O1, TeX is faster anyway. vc -O2 also works, at least with the big TeX, but doesn't seem to gain much. Convex C220 OS 8.1 cc 4.0 d.love OPT=-O1. tftopl produces bizarre indentation. DEC DECstation 3100 Ultrix 4.1 gcc 1.39 jmn Define BSD. DEC DECstation 3100 Ultrix 4.2 cc dsmith DEC DECstation 3100 Ultrix 4.2 gcc beebe With -U__STDC__ for mf/MFwindow/x11-xt.c. DEC DECstation 5000 Ultrix 4.1 #52 cc 3.1 karl Define BSD. -O makes the trip test fail. DEC DECstation 5000 Ultrix 4.0 #179 cc 2.1 kabal Define BSD. Metafont and -O don't mix. DEC DECsystem 5820 Ultrix 4.1 DEC uvax II Ultrix-32 3.1 cc karl DEC uvax 3600 Ultrix-32 3.0 gcc dick Works with glue_ratio = float; didn't try double. DEC uvax 3600 T2.1-0 (Ultrix 4.0beta) cc 1.31 karl Define BSD. -O fails. With REGFIX DEFINED. DEC uvax 5400 RISC/Ultrix4.0/rev 174 cc 2.0 dick Define BSD. With REGFIX defined. DEC Vax 11/750 4.3bsd gcc 1.37 karl DG Aviion DG/UX 4.3[01] gcc-DG-1.37 newton Define BSD. Encore Multimax 520 Umax 4.3 edh ESV3+ cc mok Harris HCX-9 HCX/UX 3.5 gcc 1.37.1 piet Define BSD, ANSI, SIGNAL_HANDLER_RETURN_TYPE=int. HP9000/[34]00 HP-UX 7.0[x] cc pfs Define _POSIX_SOURCE and HP. Set OPT=-O +Nw500 +Ns2500. Optimization loses on web2cy.c, web2cl.c, web2c.c, mf9.c, and gftodvi.c. HP9000/700 cc dsmith Define HP and SYSV. Set OPT="_Aa -D_HPUX_SOURCE" (perhaps unnecessary). TeX is OK optimized at level +O2; Metafont is OK only at +01. HP9000/7[23]0 HP-UX 8.05 cc tim Define SYSV, HP, and REGFIX. MF works with -O, but TeX only works at +O1. HP9000/800 HP-UX 7.0 cc mark dvitype breaks the optimizer. HP9000/827 HP-UX 8.02 cc corrigan OPT=-Aa -D_HPUX_SOURCE; no ANSI. HP9000/834 HP-UX 7.0 cc karl Define HP and SYSV. dvitype breaks the optimizer. HP9000/850 HP-UX 6.5 gcc 1.37 beebe dvitype breaks the optimizer. IBM 3090 beebe IBM AIX/370 cc bells Without optimization; define AIX and ANSI. IBM AIX/RT cc bells Define AIX and SIGNAL_HANDLER_RETURN_TYPE=int; with OPT="-O -G -a -Nn3000"; d IBM BSD/RT ACIS hc karl -O fails. IBM PS/2 cc bells Without optimization; define AIX, ANSI, and SIGNAL_HANDLER_RETURN_TYPE=int; the X11 support does not work. IBM RS/6000 AIX 3.1 cc karl Define AIX. IBM RS/6000 AIX 3.1 bsdcc bernd Define ANSI, not AIX. Intel 386 ESIX Rev.D gcc 1.37.1 uri Define SYSV; add -lnet, -lnsl_s, -lbsd to wlibs. Intel 386 Interactive Unix 2.2.1 gcc 1.40,cc karl cc -O fails. Intel 386 SCO Unix 3.2.0 gcc 1.37.1 wittman Intel 386 Xenix gcc klaus Define SYSV. Masscomp 5600 RTU 4.1A cc spqr But pltotf and dvitype don't work. MIPS/2000 RISC/os 4.5 cc 2.10 lgy For systype=bsd43 or systype=sysv. With -O2, the compiler breaks on ztakescaled in mf0.c; declare the integer variable `becareful' as `volatile integer becareful' to avoid this. MIPS ? RISC/os 4.52 cc 2.11 jlp With -O -Olimit 2000. Pyramid 9820 OSx 4.4c gcc 1.37, cc jmr In either the ucb or att universe; SIGNAL_HANDLER_RETURN_TYPE=int. Sequent Symmetry Dynix 3.0.12 cc morgan Metafont breaks the optimizer. Sequent Symmetry Dynix 3.0.12 gcc 1.36.93 morgan mf9.c must be compiled with cc. SGI Indigo IRIX 4.0 cc beebe Define SYSV, ANSI, _POSIX_SOURCE. Sun 2 SunOS 3.2 and 3.4 cc mackay Long lines in mf2.c and mf4.c break the compiler. Simple editing corrects this. Sun 3 Sun Unix 3.4 gcc 1.37,cc karl Sun 3 SunOS 4.0.3 gcc 1.37 beebe Sun 3 SunOS 4.1 gcc 1.39 karl Define BSD. Sun 386i SunOS 4.0.1 gcc 1.39,cc karl Define BSD. cc -O fails with Metafont: see ./PROBLEMS. Sun 4/20 SunOS 4.1 gcc 1.37 hrp Sun 4/110 Sun Unix 3.4 gcc 1.37 karl Sun 4/470 SunOS 4.1.1 gcc casterln Define BSD. Sun 4/490 SunOS 4.1 cc, gcc 1.39 karl Define BSD. Sun Sparcstation 1 SunOS 4.0.3c gcc 1.39, cc karl Define BSD. Sun sparc SunOS 4.1 cc dorab Tandy 6000 cc ken Define TANDY, REGFIX, glueratio=float, SIGNAL_HANDLER_RETURN_TYPE=int; with OPT="-Dvoid=int -DM_VOID -O"