hemphill@nrl-aic.UUCP (Gavin Hemphill) (01/07/87)
Well ISI has done it again. With version 3.07 they have added something to their crt0.o that is incompatible with the one that emacs (v 18.31) uses. Has anone figured out what the new symbol (is68020) does, and what the change to crt0.c is to allow emacs to link again? G++
gnu@hoptoad.uucp (John Gilmore) (01/18/87)
hemphill@nrl-aic.UUCP (Gavin Hemphill) writes: > Well ISI has done it again. With version 3.07 they have added > something to their crt0.o that is incompatible with the one > that emacs (v 18.31) uses. Has anone figured out what the new > symbol (is68020) does, and what the change to crt0.c is to allow > emacs to link again? I find this a very arrogant point of view. Gavin is complaining that ISI has changed part of the unsupported, undocumented machinery down in the guts of the system. "They've done it again." What, did he want them to change the DOCUMENTED part of the system on him instead? Or just stop working on their software? I'd say it's Gavin's problem if he is depending on this kind of stuff. Another point of view is that it's GNU Emacs's problem. GNU Emacs is a nice program, but it was not written to be portable -- it was written to run under GNU, or maybe under Unix until we get GNU finished. Somehow I doubt that under GNU, the guts of crt0.s will be documented and reliable from release to release and hardware to hardware, but that was Richard's point of view last time I asked. When I first tried to bring up GNU Emacs and discovered that it replaced malloc() with one that used an initialization call, crt0.o with its own custom version, required an alloca() which many systems can't support, and replaces a few other system subroutines with incompatible local versions (with the same names as system routines), I was incensed and stopped messing with it. Since then, somebody's hacked a crt0 for Sun-3's, and somebody else hacked a portable alloca() that only requires a couple of source changes to Emacs, so Emacs runs on my machine, but I haven't used it or tried to maintain it. Besides, its mhe is worse than the original one for Gosling Emacs, and that and srccom are all I use emacs for anyway. [I regularly use Unix utilities that are encrusted with horrible code. Maybe part of the drawback of getting the source code is that you can get so disgusted with its internals that you won't use it because you know it's not maintainable, even if it's useful.] -- John Gilmore {sun,ptsfa,lll-crg,ihnp4}!hoptoad!gnu gnu@ingres.berkeley.edu /* No comment */