Path: utzoo!utgpu!news-server.csri.toronto.edu!mailrus!umich!samsung!zaphod.mps.ohio-state.edu!mips!orac!beacker From: beacker@mips.COM (Bradley Eacker) Newsgroups: comp.sys.mips Subject: Re: kermit and cc -i Keywords: kermit Message-ID: <40048@mips.mips.COM> Date: 11 Jul 90 01:50:33 GMT References: kermit Sender: news@mips.COM Reply-To: beacker@mips.COM (Bradley Eacker) Followup-To: beacker@mips.com Organization: MIPS Computer Systems, Sunnyvale, CA Lines: 42 William Bader writes: >A while back, someone posted that they had to build kermit by building >wart (a lex-workalike) with -systype bsd43 and the rest of kermit with >-systype sysv. I played around with compile options, and it seems to me >that the culprit is the '-i' flag in the kermit makefile for sysv kermits. >I have two questions: >1) Does the Mips compiler (or linker) have a problem with '-i', or >is this some subtle bug in ckwart.c that only shows up on Mips computers? >ckwart.c lints cleanly. I put debug code in the main program, and some >in a function (setstate() ?), and the debug code in the function printed >BEFORE the debug code at the top of the main program. How can this happen? The -i flag has a different meaning in the mips linker. It says to put the code into the initialization section. This is not the separate I & D space flag that is used in other compilers. ld(1) explains this in better detail. >2) If '-systype bsd43' simulates a BSD 4.3 environment, why won't it >compile the bsd version of kermit? Some routines in kermit, like msleep() >in ckutio.c run much more efficiently in a bsd environment. The current Makefile that is shipped with the 4.50 environment has the following defines for the compiler flags: MAXOPT =-O1 INCLUDE = CFLAGS =$(MAXOPT) $(INCLUDE) $(ECFLAGS) -DSVR3 -DUXIII -DDEBUG -DTLOG LNKFLAGS= These are the necessary flags to build a sysV kermit which is what the main basis is for RISCos. You may be able to build a bsd version by substituting "-systype bsd43 -DBSD4" inplace of "-DSVR3 -DUXIII" I have not tried this since the sysV version seems to function in a reasonable fashion. Brad Eacker (beacker@mips.com)