Hi CHARLES.

09-Aug-03 21:58:00, CHARLES ANGELICH wrote to JASEN BETTS


 CA> 1237c94464c2 c_echo

 CA> Hello Jasen -

 CA>>>>> I haven't used getopt() for some time now but I have often
 CA>>>>> wondered what the reasons are that most methods of
parsing the
 CA>>>>> command line want to use switch characters such as
"/" or "-"?

 JB>>>> usually to separate switches (ie commands) from filenames or
 JB>>>> other literals

 CA>>> With argc available I don't see the need for the switch chars.
 CA>>> Order of entry should be sufficient.

 JB>> for trivial tools it often is.

 CA> To be honest I only write trivial tools, the more trivial the
 CA> better. I like to think I'm following in the *nix model of just
 CA> piping from one to the other but I detect a good deal of inertia
 CA> when even thinking about taking on writing a complete app with
 CA> user-interface etc. I'm seriously considering Steve Gibson's offer
 CA> of a template for writing Windows apps but it's all ASM and I
 CA> would have to install a Windows compiler/assembler to use it.
 CA> Getting lazy in my old age. ;-)

 JB>> try designing something that can do,everything the dos copy
 JB>> command can do (well the things it's documented as being capable
 JB>> of) without it using any switches...

 CA> There are so many clones of "copy" I never thought to write
 CA> another one but I have written a dropin replacement for "sort" in
 CA> all C code.

same.... limited to available meory but atleast it uses an Order(n.log(n))
sort algorithm (qsort() actually) (which is a vast improvemnt on the
n-squared algorithm in the dos one.

   I wrote a qsort program in qbasic and on a file of 2000 random short
   words the interpreted sort.bas program could actiually beat the dos
   sort.exe

 CA> I could fiddle with that one and see if I could remove
 CA> the switch characters as an exercise but since it is intended to
 CA> pipe input and output I suspect that would be far too easy. :-)

the dos version takes an optional filename... could prove challenging

 CA> Yes, I use AWK occasionally and like it (recently downloaded
 CA> Kernighan's W95 executable for AWK) but putting the code into the
 CA> command line is somewhat tedious (for me) with the need to escape
 CA> characters etc. Starts to look like PERL code eventually and I
 CA> don't like PERL. :-)

yeah, code in the command-line is more a unix thing where there are fewer
limits on command line length.

I'd post the source for the sort but it's full of turbo-C isms and dos
specific stuff.
basically all it did was read the file into memory then make and index of
the lines in an array of pointers while null-terminating each one. Then it
qsort()ed the index using a compare function based on stricmp and finally
used a for loop to go theough the index and print out each line one at a
time.

 -=> Bye <=-



---