On (30 Jul 96) Roy McNeill wrote to Paul Edwards...



 RM> Hi Paul



 RM> I'm having trouble getting Indent (from Indentx.zip dated 10-6-93,



 PE> Grab INDENTPE.ZIP instead, much better.



 RM> Will do.



I did, and it's a mess. The .exe file in the zip file says it can't

find emx.exe, and terminates itself, so I had a go at compiling the

source. No chance. Most of the functions are not declared in header

files, so calls from other files fail. There are two different

calling techniques used for the function diag() - (int, char*, int,

int) and (int, char*, char*) - the compiler just loved that. A heap

of functions requiring stdio.h, stdlib.h, string.h, fcntl.h, and

io.h are used, although none of those headers are d.



I finally got it to compile after nearly a day, with only 76

warnings (eg heaps of "significant digits may be lost in

conversion", and a couple of very suspect "Ambiguous operators need

parentheses"), and even now it don't go proper. A number of lines

it just fails to recognize, including every line starting with #

except #else and #endif, and "case x:" statements - it just deletes

them and writes a single P (correctly indented) instead. I can't

get it to keep the "int" and the "foo(void)" in
"int foo(void)" on

the same line, no matter what parameters I use.



It brought back some memories, though. I haven't seen the quaint

usage



  char *memcpy (destaddr, srcaddr, len)

       char *destaddr;

       char *srcaddr;

       int len;

  { 



since I started with Turbo C v1 back in when Noah was writing

Basic.



And there's one cute little bit where an int pointer member of a

structure is recast as a pointer to an int function, and is then

called... Rather tricky, although the compiler gets a bit red in

the face about it. I'll run it through the debugger sometime and

see if it actually works.







However, I think I'll stick to manual indenting. Ta anyway.



Cheers



--- PPoint 1.88