/* * TIPC.C - TI Professional and compatible terminal driver for MicroEmacs. * This version by Ronald Lepine, Moderator - TI Conference * Byte Information Exchange. * * BIX ID. ronlepine * Programers Room - Ronald Lepine (317) 742-5533 2400 baud * AT&T: (302) 836-8398 * US Mail: * Ron Lepine * 434 Shai Circle * Bear, DE 19701-3604 * * Adapted from code by Danial Lawrence for the IBM-PC. * * February 1993, David G. Holm (email address: dgh@bix.com) * Updated to work with new TERM structure in 3.12 BETA 1. * Updated to work with Borland's C/C++ compilers. * * The routines in this file provide support for the TI-PC and other * compatible terminals. It truely writes directly to the screen RAM to do * screen output. It compiles into nothing if not a TI-PC driver. * * This driver for the TIPC is not the same one that Daniel Lawrence * distributes. He uses TI's DSRs for most functions as you can see by * looking at the driver in ue311c.arc. Dan's reverse video * also does not work. But do not blame him as I'm sure he does not * have a TIPC to test Emacs on and he *is* supportting the TIPC. * [Well, as you can see, it is the one I distribute now! - DAN] * This driver uses direct writes to screen memory except for the message * line like the IBM driver. In fact I keep TI and IBM versions around by * having my make file replace estruct.h with one that has a IBMPC defined * only when I compile IBMPC.C and then switching back to estruct.h with * TIPC defined. Its last step is of course linking in the correct object * module for each exe. * * You can define any buffer to be reverse video by setting the * foreground to "BLACK" and the background to any other color * such as "cyan" to get Black on Cyan text. There is no way to get * colors such as 'White on blue' on the TIPC unless it has a * least a 1-plane graphics card in it. If it does have a graphics * card installed the code to change overall background * color is tied to the $palette variable with the same values as * TI BASIC and the Tech Reference manual section 3.5.2 as well as * being the #define for the color here without the enable attribute. * The text mode forground/background code still works even when you * use a $palette backgound giving you a wide range of what can be * done even though some combinations don't really work too well. * * The only reason that there are any '#if color' directives is because of * the need to maintain structure alignments. This is because TI's mono * is really 8 shades instead of 8 colors and all attributes remain the * same. You can have color and mono attached at the same time and * display on both without any program support or hardware changes. So * you should define color. It *will* make a difference in other areas of * the MicroEmacs code. You also do not have to do anything special for * the palette code because writing to a nonexistant graphics plane on a * TI causes *no* ill effects (at least according to the tech manual). * * Note that if you recieve MicroEmacs for the TIPC compiled by me * it is compiled/linked so that wildcard file names on the command * line are automagicly expanded in case you want to work with several * similarly named files. * * Changes to the other files to support the the TIPC MEMAPed * version compiled under Zortech C/C++ 1.02 to 2.10 follow. * BTW - The version I used is 2.18 but the earlier versions should * still be similar since I have compiled previous versions under whatever * was current at the time and even the previous ZTC version sometimes. * * [I simply added Ron's #ifdef changes to the master sources - Dan] * * I once thought there were three ways to correct the problems with * DSR writes to the *25th* line, I now know some don't work. * One was to check the start register before all screen writes. After * some investigation and testing I found that the 6545A Start of screen * registers are write only. So there is not way to correct cursor * postioning and line position whenever DOS has changed the register. * Changing the start of screen pointer to offset 0 before screen writes * also didn't solve the problem completely. As soon as you touched a key * after the error you ended up with the screen properly postioned, but * you had the error message on the screen until that line was updated so * you still ended up using a ^L to correct the screen. * Two is to capture critical errors (mlwrite and errors are the sources * of DSR writes to the 25th line) and handle them myself (asm is * only clean way) and now looks like the only way it is possible but must * be used in conjuction with a modified method three too. * Three was change writes to the message line to direct screen writes * and initialize the screen so that a critical error will write to the * text portion of the screen where any repaints would clear it. The same * problems cropped up that did in number one where you needed to do a ^L * to totally clean up the screen if the error occured while on the message * line anyway so.... * Therefore a combination of two and three is most likely the best * way if it can be done since TI does not recommend mixing DSR and direct * writes to screen (if I could read the 6545 registers 0Bh and 0Ch I * could manage it though). This is because the DSR writes *can* change * the screen start register. The problem with a 2+3 solution is mainly * that the extra code to handle single character writes, directly * controlling the cursor though the CRTC cursor registers, and the * intercept/write/read error code handling bloats the code a bit and * coding in anything but assembly language (which then needs to be * interfaced to each C compiler that might be in use) is the only reasonable * way. I just came came to the conclusion that it wasn't worth it since * a ^L will clear things up anyway after the error is corrected. * * 12/29/89 - * Changed lnptr, sline, and scptr to char/char*'s because TI writes * only a character at a screen position and IBM a char + attribute. TI's * attributes are kept up by the hardware except when changed. So we only * write a char at a time, not an int like the IBM code did. This also * corrected a minor bug with the message line eeol's which is why I * looked at the driver code again. * * 12/30/89 - * added the capability of setting the background on machines with * graphics by setting the palette string with a number from 0 to 7. * you do this with ^XA $palette (note small letters) and a value. * Values >= 8 will work since I do modulo division on the value * but results are the same as the color values in TI BASIC. The * code, as always, works on all TI's but will not produce a change * on non-graphics equiped TI's. This code also does not need a color * monitor to work since a TI doesn't know what kind of monitor is * attached and may even have color and mono attached at once. * * With the code set up so you can still do reverse video on any TI * by defining BLACK and yellow (or any other background color) you * have a large variety of posible screen setups. So experiment. * * 12/31/89 to 1/7/90 - experimented with various methods of eliminating * problems with mixed DSR/Direct_Memory writes. Decided not to really * attempt a solution at this time. Write Ron Lepine if you have * any feelings/suggestions on this problem, they are welcome. * * 1-9-90 * 3.10.c(Beta) New IBMPC reverse video code added to TIPC driver. * * Addional changes to 3.10.c(beta) to get it to compile with Zortech C/C++ * **** NOTE THESE CHANGES ARE NO LONGER NEEDED FOR ZTC 2.10 ***** * isearch line 172/5, 191 - change lines to * * if (kfunc == &forwsearch || kfunc == &forwhunt || * kfunc == &backsearch || kfunc == &backhunt) * { * dir = (kfunc == &backsearch || kfunc == &backhunt)? * REVERSE: FORWARD; * * 5-26-90 * Updated first part of this file to reflect 3.11beta line numbers * while compiling 3.11 beta under Zortech C/C++ 2.1. * * 6-04-91 * Added last eight needed values to ctrans[] so standard emacs.rc file * now works. */ #define termdef 1 /* don't define "term" external */ #include #include "estruct.h" #include "eproto.h" #include "edef.h" #include "elang.h" #if TIPC #define NROW 25 /* Screen height. */ #define NCOL 80 /* Screen width */ #define MARGIN 8 /* size of minimim margin and */ #define SCRSIZ 64 /* scroll size for extended lines */ #define NPAUSE 200 /* # times thru update to pause */ #define BEL 0x07 /* BEL character. */ #define ESC 0x1B /* ESC character. */ #define SPACE 32 /* space character */ #define SCADD 0xDE000000L /* address of screen RAM */ #define ATTRADD 0xDE001800L /* Address for attribute latch */ #define BLUE_PLANE 0xC0000000L /* Address Blue graphics plane */ #define RED_PLANE 0xC8000000L /* Address Red graphics plane */ #define GREEN_PLANE 0xD0000000L /* Address Green graphics plane */ #define CRTC_REG 0xDF810000L /* 6545 CRT Controller access addr */ #define CHAR_ENABLE 0x08 /* TI attribute to show char */ #define TI_REVERSE 0x10 /* TI attribute to reverse char */ #define BLACK 0+CHAR_ENABLE /* TI attribute for Black */ #define BLUE 1+CHAR_ENABLE /* TI attribute for Blue */ #define RED 2+CHAR_ENABLE /* TI attribute for Red */ #define MAGENTA 3+CHAR_ENABLE /* TI attribute for Magenta */ #define GREEN 4+CHAR_ENABLE /* TI attribute for Green */ #define CYAN 5+CHAR_ENABLE /* TI attribute for Cyan */ #define YELLOW 6+CHAR_ENABLE /* TI attribute for Yellow */ #define WHITE 7+CHAR_ENABLE /* TI attribute for White */ PASCAL NEAR ttopen(); /* Forward references. */ PASCAL NEAR ttgetc(); PASCAL NEAR ttputc(); PASCAL NEAR ttflush(); PASCAL NEAR ttclose(); PASCAL NEAR timove(); PASCAL NEAR tieeol(); PASCAL NEAR tieeop(); PASCAL NEAR tibeep(); PASCAL NEAR tiopen(); PASCAL NEAR tikopen(); PASCAL NEAR tirev(); PASCAL NEAR ticres(); PASCAL NEAR ticlose(); PASCAL NEAR tikclose(); PASCAL NEAR tiputc(); PASCAL NEAR tifcol(); PASCAL NEAR tibcol(); PASCAL NEAR scinit(); int revflag = FALSE; /* are we currently in rev video? */ int cfcolor = -1; /* current forground color */ int cbcolor = -1; /* current background color */ int ctrans[] = /* ANSI to TI color translation table */ {BLACK, RED, GREEN, YELLOW, BLUE, MAGENTA, CYAN, WHITE, BLACK, RED, GREEN, YELLOW, BLUE, MAGENTA, CYAN, WHITE}; char *scptr[NROW]; /* pointer to screen lines-8 bit on ti * char sline[NCOL]; /* screen line image */ #if __TURBOC__ char far * scadd = SCADD; /* address of screen ram */ char far * attradd = ATTRADD; /* address of attribute latch */ char far * blue_plane = BLUE_PLANE; /* address of blue graphics plane */ char far * red_plane = RED_PLANE; /* address of red graphics plane */ char far * green_plane = GREEN_PLANE; /* address of green graphics plane */ char far * crtc_reg = CRTC_REG; /* 6545 CRT controller address */ #else long scadd = SCADD; /* address of screen ram */ long attradd = ATTRADD; /* address of attribute latch */ long blue_plane = BLUE_PLANE; /* address of blue graphics plane */ long red_plane = RED_PLANE; /* address of red graphics plane */ long green_plane = GREEN_PLANE; /* address of green graphics plane */ long crtc_reg = CRTC_REG; /* 6545 CRT controller address */ #endif /* * Standard terminal interface dispatch table. Most of the fields point into * "termio" code. */ TERM term = { NROW-1, NROW-1, NCOL, NCOL, 0, 0, MARGIN, SCRSIZ, NPAUSE, tiopen, ticlose, tikopen, tikclose, ttgetc, tiputc, ttflush, timove, tieeol, tieeop, tieeop, tibeep, tirev, ticres #if COLOR , tifcol, tibcol #endif }; extern union REGS rg; PASCAL NEAR tifcol(color) /* set the current output color */ int color; /* color to set */ { cfcolor = ctrans[color]; } PASCAL NEAR tibcol(color) /* set the current background color */ /* color to set */ int color; { cbcolor = ctrans[color]; } PASCAL NEAR timove(row, col) int row; int col; { rg.h.ah = 2; /* set cursor position function code */ rg.h.dh = col; rg.h.dl = row; int86(0x49, &rg, &rg); } PASCAL NEAR tieeol() /* erase to the end of the line */ { int i; /* loop variable */ char *lnptr; /* pointer to the destination line */ int ccol; /* current column cursor lives */ int crow; /* current row cursor lives */ /* find the current cursor position */ rg.h.ah = 3; /* read cursor position function code */ int86(0x49, &rg, &rg); ccol = rg.h.dh; /* record current column */ crow = rg.h.dl; /* and row */ lnptr = &sline[0]; /* set things up */ for (i=0; i < term.t_ncol; i++) *lnptr++ = SPACE; _fmemset(attradd, cfcolor, 1); /* write current attrs to latc movmem(&sline[0], scptr[crow]+ccol, term.t_ncol-ccol); } PASCAL NEAR tiputc(ch) /* put a character at the current position */ /* in the current colors */ int ch; { _fmemset(attradd, cfcolor, 1); /* write current attrs to latch */ rg.h.ah = 0x0E; /* write char to screen with DSR */ rg.h.al = ch; int86(0x49, &rg, &rg); } PASCAL NEAR tieeop() /* Actually a clear screen */ { rg.h.ah = 0x13; /* Clear Text Screen, Home Cursor and */ int86(0x49, &rg, &rg); /* Set screen start register to 0 */ } PASCAL NEAR tirev(state) /* change reverse video state */ int state; /* TRUE = reverse, FALSE = normal */ { revflag = state; } PASCAL NEAR ticres() /* Change screen resolution. Should we add the */ /* the 720x350 mode besides the normal 720x300 */ /* mode for those who can handle it? It really */ /* gains us nothing since we don't gain lines */ /* and the code works when started in that res. */ /* Let me (Ron Lepine) know what you think */ { return(TRUE); } PASCAL NEAR tibeep() { bdos(6, BEL, 0); } PASCAL NEAR tiopen() { strcpy(sres, "NORMAL"); revexist = TRUE; revflag = FALSE; scinit(); ttopen(); } PASCAL NEAR tikopen() { /* Does nothing */ } PASCAL NEAR ticlose() { _fmemset(attradd, WHITE, 1); /* write normal attrbute to latch */ /* Makes sure we return with a normal color */ /* and not reverse video or such. Attribute */ /* Latch will otherwise hold last color and */ /* attributes it wrote to the screen, which in */ /* some cases is reverse video. */ } PASCAL NEAR tikclose() { _fmemset(attradd, WHITE, 1); /* write normal attrbute to latch */ /* Dan doesn't close the the terminal when shelling */ /* This ensures the shell starts with a normal color */ /* the same way ticlose does without changing */ /* *any* other source file. The only real reason we */ /* place it here is to change as few files as posible */ } PASCAL NEAR scinit() /* initialize the screen head pointers to */ /* logical screen */ { union { long laddr; /* long form of address */ char *paddr; /* pointer form of address */ } addr; char i; tieeop(); /* set logical = physical start of screen */ /* initialize The screen pointer array */ for (i = 0; i < NROW; i++) { addr.laddr = scadd + (long)(NCOL * i); scptr[i] = addr.paddr; } } PASCAL NEAR scwrite(int row, char *outstr, int forg, int bacg, int revleft, int /* write a line out */ { char *lnptr; /* Pointer to the destination line */ char i; /* Write the attribute byte to latch and setup the screen pointer */ /* If forg == 0 then you can change the back ground to color. TI */ /* won't let you use two colors for forground and background unless */ /* you have graphics and set the graphics color to your background */ /* This is enabled in this driver though the $palette variable */ if ((forg != 0) & (!revflag)) { forg = ctrans[forg]; _fmemset(attradd, forg, 1); }else if ((forg != 0) & (revflag)) { forg = ctrans[forg] + TI_REVERSE; _fmemset(attradd, forg, 1); }else{ bacg = ctrans[bacg] + TI_REVERSE; _fmemset(attradd, bacg, 1); } lnptr = &sline[0]; for (i=0; i