bobmon@silver.bacs.indiana.edu (outsider) (03/22/88)
Wow! A program posting to comp.binaries.ibm.pc! I was so excited that I rushed right home and recompiled it. Well, okay, I recompiled it because I wanted to change its default insert-mode to ON. But here's the quirky thing that I don't understand -- the included executable was about 5K bytes long. I changed two occurrences of "FALSE" to "TRUE", ran it through tcmake (the makefile clearly expects Turbo C, fortunately for me), and I get an executable that's about 9K bytes long. I applied patches that were posted at the same time -- same size. So what am I doing wrong? More details: this is history v1.4R. I notice that the modified c0.c that was supplied with v1.3 is missing -- should I be using that? confusedly, bobmon
nelson@sun.soe.clarkson.edu (Russ Nelson) (03/23/88)
In article <1239@silver.bacs.indiana.edu> bobmon@silver.UUCP (outsider) writes: >Wow! A program posting to comp.binaries.ibm.pc! Yes, I was pleased to increase to signal/noise ration there :-) >But here's the quirky thing that I don't understand ... >[after compiling] I get an executable that's about 9K bytes long. Well, I had compiled it at home with my TC 1.0. When I got into work after the weekend and compiled it with TC 1.5, there it was at 9K. Turns out that putch() in TC 1.5 sucks in Borland's new text windowing package. I'll post a new version after I add a configuration option, since people who have extended keyboards have trouble typing ^J for filename completion. Also for people who like insert mode as the default :-) >More details: this is history v1.4R. I notice that the modified c0.c that >was supplied with v1.3 is missing -- should I be using that? No. I have a better scheme for writing a TSR. I simply use malloc() (actually sbrk()) to grab whatever memory I need, then do a _SP = sbrk(1024); and then keep((_SP + 15) >>4); This way, I can avoid modifying c0.asm, which is undesirable. -- -russ AT&T: (315)268-6591 BITNET: NELSON@CLUTX Internet: nelson@clutx.clarkson.edu GEnie: BH01 Compu$erve: 70441,205
bobmon@silver.bacs.indiana.edu (outsider) (03/23/88)
nelson@sun.soe.clarkson.edu.UUCP (Russ Nelson) writes: > >>But here's the quirky thing that I don't understand ... >>[after compiling] I get an executable that's about 9K bytes long. > >Well, I had compiled it at home with my TC 1.0. When I got into work after >the weekend and compiled it with TC 1.5, there it was at 9K. Turns out >that putch() in TC 1.5 sucks in Borland's new text windowing package. I'll >post a new version after I add a configuration option, since people who have >extended keyboards have trouble typing ^J for filename completion. Also >for people who like insert mode as the default :-) > I get it -- Microsoft starts out with <insert your favorite insult here>, and asymptotically approaches a good product; Borland starts out with a good product, and asymptotically approaches Microsoft. Proof: Turbo C is at v1.5 Microsoft C is at v5.1 Q.E.D. For what it's worth, BTW, and all that jazz: I don't have an extended keyboard (640K keycodes is enough...) but, using Fansi-Console, I can generate a Linefeed (0x0A) with a ctrl-RETURN. I have no idea whether this works with nansi/ansi/no-driver, as I haven't taken the time to reboot with a different config.sys. (Oh, Z-158 keyboard if that matters.)