wjc@ho7cad.UUCP (06/20/87)
In article <1362@super.upenn.edu.upenn.edu>, brant@linc.cis.upenn.edu.UUCP writes: > Does anyone out there regularly port the latest distribution of > GNU Emacs (currently 18.46) to the PC 7300/3B1? Has anyone ever > tried it? Is it hard or easy? If no one's done it, I'll probably > try. I compiled 18.41 (using the UNIXpc 3.5 compiler, etc) and it worked without a hitch (there is a 7300.h, but it seems to be for the old pre3.5 software). (1MB RAM, don't know if it matters.) When I say it worked fine, I mean without the "unexec()" option. So, all that lisp gets loaded up each time. Takes about 2 minutes. I'm sure glad that GNUemacs is a program you stay in for a while, but it makes it useless as an escape editor for mailers and the like (obviously, you can solve the mail problems by using GNUemacs mail stuff, but the more general problem continues). Anyhow, I asked a couple weeks ago in comp.emacs if anyone had gotten GNUemacs to run on the 7300 with the preloading. No response. I fiddled with it for a couple days and gave up. Not interested in learning 68010 assembler just now. If anyone has gotten this to work, I'd like to hear about it. Also, if anyone is interested, I'd be happy to pass along the modified congifuation headers, although it's pretty easy to do. Bill Carpenter (AT&T gateways)!ho5cad!wjc HO 1L-410, (201)949-8392
wjc@ho5cad.ATT.COM (06/20/87)
In article <175@ho7cad.ATT.COM>, wjc@ho5cad.ATT.COM writes: > > I compiled 18.41 (using the UNIXpc 3.5 compiler, etc) and it worked > without a hitch (there is a 7300.h, but it seems to be for the old > pre3.5 software). (1MB RAM, don't know if it matters.) Sorry, I meant ___2 MB RAM___. I don't know if it even runs in less, was my point. BTW, after you throw away nonessentials, compress the source (C and lisp), etc, it costs about 5 MB to keep this on your machine. I guess you could recover another meg if you got rid of the stuff you compressed. Bill Carpenter (AT&T gateways)!ho5cad!wjc HO 1L-410, (201)949-8392
jkw@mtung.ATT.COM (Kevin Wanner) (06/22/87)
You can can get unexec to work correctly by including the -N option to the LD_MACHINE_SWITCH in m-7300.h This forces the loader to load the data right after text, which is what unexec expects, e.g. the last three lines of my m-7300.h file are: #define LIBS_TERMCAP -lcurses -ltermcap #define LD_MACHINE_SWITCH -s -N #undef TERMINFO Good Luck! Kevin Wanner [ihnp4|allegra|akgua]!mtung!jkw (201)957-2704
Karl.Kleinpaste@cbstr1.att.com (06/23/87)
Posting-Front-End: GNU Emacs 18.47.3 of Sat Jun 20 1987 on cbstr1 (usg-unix-v) jkw@mtung.ATT.COM writes: > #define LD_MACHINE_SWITCH -s -N Be aware that use of -N causes SysV's ld(1) to create the binary as a single, writable (hence unshared) segment. On a 7300, where (one hopes) you are all alone by yourself with only a single emacs running, that's OK, but as soon as you need a second simultaneous incarnation of emacs, you're going to find yourself in deep performance trouble. Karl
wjc@ho5cad.ATT.COM (06/24/87)
Posting-Front-End: GNU Emacs 18.41.6 of Tue Jun 16 1987 on ho5cad (usg-unix-v) In article <961@mtung.ATT.COM> jkw@mtung.ATT.COM (Kevin Wanner) writes: > You can can get unexec to work correctly by including the > -N option to the LD_MACHINE_SWITCH in m-7300.h This forces the > loader to load the data right after text, which is what unexec > expects, e.g. the last three lines of my m-7300.h file are: > > #define LIBS_TERMCAP -lcurses -ltermcap > #define LD_MACHINE_SWITCH -s -N > #undef TERMINFO > > Good Luck! > Kevin Wanner > [ihnp4|allegra|akgua]!mtung!jkw > (201)957-2704 > Well, that's finally answered for me. Thanks! BTW, the #define is actually LD_SWITCH_MACHINE rather than the vice versa shown above. The "m-7300.h" file is set up for the pre3.5, preflexname development environemnt on the UNIXpc. If you have the 3.5 development utilities, you might be interested in these changes I made in "config.h": #include "s-usg5-0.h" #undef ADDR_CORRECT /* simply to suppress "redefined" msg */ #include "m-7300.h" #undef SHORTNAMES /* 3.5 has flexnames */ #define LD_SWITCH_MACHINE -s -N /* as seen above */ #define TERMINFO /* 3.5 has true terminfo */ #define SWITCH_ENUM_BUG /* Some "bug"! */