1Current Xspread supporter: 2Included below 3 1) High level list of change from Xspread3.1.1 to this version. 4 2) A detailed list of changes to xspread. 5 3) A note I sent earlier. 6 7 8 -Jeff Buhrt 9 Ameritech Services, Inc. 10 317-265-7426 11 I-net: jeff@appf.ci.in.ameritech.com 12 13*********************** 14Major things that have changed (Xspread3.1.1 to this version) 15 -X & Curses (text/sc) mode will automatically be selected based on 16 availability of X libs, $DISPLAY, etc. 17 -X can be deselected (curses used) by adding '-X' to xspread 18 -Gnu autoconfig creates a 'configure' script that automatically 19 searchs for system features vs hand updating configs. 20 -The Quick Reference card, scqref (formerly a separate program), 21 which xspread has never printed, can be printed by: 22 xspread -q # text format output 23 xspread -qt # troff -ms format output 24 xspread -h # prints usage from help.c vs version.c 25 (allows qref to use same data) 26 -Addition of functions: 27 evalall, @irr, read (read text strings) 28 [copy can now copy to a cell (it figures out the size needed)] 29 -Corrected/added 30 -enhanced vi support 31 -Framemaker table support (file suffix/format) 32 -Get/Put/etc. now consistant on filling in with defaults 33 -cleaned code for portability & limiting compiler warnings 34 -Makefile now builds 'pxspread' (the name carried from 'psc') 35 -Psc will take ascii files and try to convert them into 36 xspread data. 37 -Tested on AIX 3.2.5 & 4.1.3, AT&T SysV R3.2.3, Linux slackware 1.59 38 39*********************** 40CHANGES: 411) ierr = 1; // means no error ? 42 -ierr = 0 makes more sense for no error 43 44other: (grep JEFFB) 45 -format.c: 0 fill and precision might not always work together 46 -graphic_main.c: ~line209 47 note: case 0 will never be called 48 49graphic_main.c: 50 -~line950 graphic_skipscale() was never 51 -NULL==0 (most of the time) thus case 0 is skipped 52 -I moved: graphic_init(), graphic_read_defn(), graphic_write_defn() 53 to the bottom, they are needed whether we have X11 or not 54 55cleaned up a few places where: 56 -NULL was used vs the integer 0 57 -if (pointer = funt()) -> if ((pointer = funt()) != NULL) 58 because is (x = 0) intended or (x == 0) 59 -gcc suggests adding extra '()' as a minimum, the 60 most readable is to be explict in the compare 61 62moved message(char *) into util.c to be next to error(char *) which 63 -does almost the same thing (different display line and a sleep). 64 -scXstuff.c isn't used in !DOINGX mode 65 66added missing function prototypes 67 -'function PROTO((arg1, arg2, ...))' is now used vs the old __STDC__ 68 listing each function twice (once with and once without args). 69 -PROTO() is eather '#define PROTO(x) x' [when we are ansi and can 70 handle function arguments in prototypes] 71 or '#define PROTO(x)' [when we are traditional K&R] 72 73Things that are shared should be defined in a common header included by 74both files so the compiler can cross check. (extension of above) 75(added missing) 76Ex: the struct's: freeents & freeenodes 77 functions 78 79added '{}' around the structure initializer pairs in [es]res.sed 80 -They are used to produce: experres.h and statres.h respectively, 81 which are included by lex.c. 82 -This allows gram.y to have more comments on what things mean. 83 84Curses: 85 -Fixed a few curses problems that DOINGX didn't take into account, 86 including but not limited to: 87 -move(x,y) (incorrect) was used vs move(y,x) in clearlines() 88 -Added HAVE_X*: now supports curses & X in the same executable 89 -DOINGX was dropped and replaced... 90 91screen.c 92 -in update() 'char field[]' didn't need to be static 93 -added safety checking of default screen size to make sure 94 curses xspread could run w/ a file as stdin 95 [this was very stumping until I realized xspread never left update(), 96 it was constantly try to place the center row/col between two 97 impossible limits] 98 99get/put default directories 100 -auto adding $HOME to the path on a get/put makes life harder, 101 you have to backspace over/edit out the expanded path, if 102 people store everything in their $HOME maybe. What if I am 103 in a directory path hundreds of chars deep (maybe?), the default 104 should be where I am. 105 -a possiblity is a .xspreadrc that has personal preferences... 106 FIXED BY: using what default said before and some very nice 107 code that was sent in after Sc6.28 (but never used in a released 108 version of sc). 109 110Other 111 -name change: error() -> scerror() for ansi/gcc/AIX conflict 112 -eres.sed/sres.sed now put {} each line (struct element) and 113 handle comments in gram.y (detail above) 114 -The W command will give as much space as needed to a label in 115 the last column being written, so can be used with r to read and 116 write arb line length text files. 117 -added FRAME table type suffix. 118 -Centralized table suffix generation. 119 120 121gram.y 122 -added comments (merged from my local version) 123% -added the finacial function IRR -Internal Rate of Return 124 '@irr()' computes internal-rate-of-return over a range 125 containing a cash flow. 126% -added S_READ -read a string delimited file to a range 127 'r [range_or_var]' command which reads the contents of a file 128 as strings into a range or starting at a cell. 129% -added S_EVALALL 130 command 'evalall' for use in scripts (the default is 131 not to to any evaluation while reading a file) 132% -S_COPY to a var, ex: copy to a50 from a10:f30 133 (make the destination conform to the source) 134 135%-needs added to doc/* files 136 137vi.c 138 -support for D and C in edit mode, as well as c$, d$, c0, d0. 139 Man and help pages. 140 -fixed incorrect interpretation of cl command at end of line 141 142 143TODO: 144auto-config PAGER (use env) if fails use internal 145CRYPT same as above... 146SIGTSTP causes an endless loop (curses on Linux) 147 148CHECK: 149 -reg*c*mp don't seem to work same way 150 -probe for: 151 -HAVE_X11* 152 153*********** 154I just noticed the [somewhat] new Xspread3.x code... 155 156Some notes on the new Xspread3.1.1: 1571) In the C code: '\0' should be used vs NULL to end strings 158 NULL is a pointer, '\0' end of string, 0 an integer 159 (CDC even had a few UNIX machines where: NULL != '\0' != 0) 1602) Makefile: $(OBJS) should not include gram.o so gram.o doesn't 161 try to be built by 2 different make targets 162 1633) AIX3.2 needs -lPW for regcmp/regex 1644) Per the TODO file: 165-TODO- 166> BTW, there is a serious bug concerning functions that return strings 167> (eg @DATE, @SUBSTR): it says "invalid numeric function" or somesuch. 168> This means that eval() instead of seval() is being called for those 169> functions, but I couldn't discover why. At present, you can't use 170> functions that return strings. 171 172 Strings functions are assigned the string part of a cell. 173 Use: label, leftstring, rightstring (verbs in gram.y that call slet()) 174 vs let (which calls let()) 175 176 I always thought that something could be done to make it automatic. 177 a) ie: [xspread saying this] 'I know that he said assign this to the 178 numeric part, but what is given is a string therefor I will just 179 do a 'label' or such vs the let. 180 b) rethink the numeric/string idea of cells [note: major change] 1815) I will look into what was broken with the curses, aka 'sc' part, of xspread. 1826) I have little to no time in the last few years for support of 'sc' and 183 would love to turn over the maintainership. Really Xspread is becoming 184 stable enough, as long as the text/curses part were supported, just 185 to merge officially (become the same product). 186-TODO- 187> Currently this is xspread.. One day it will be incorporated 188> with SC. SC and xspread will be the same program. 189> I've also noticed that sc.c could not compile if DOINGX is not defined 190> due to the use of userfont which is declared only when DOINGX is 191> defined. Some of the other defines also still get screwed up. 192 1937) I will answer any questions as needed, etc. 194 195 196