Re: C-programmer needs Forth advice

Jeff M. wrote: 
First off, macros are great. These macros are just plain bad. You may
use them. Fine. (Why should I care?) However, that is a gross misuse of
a language feature.


So, care to tell me *why* or are you just going to repeat yourself?

Second, there is no such thing as a "local macro". Macros in C are
handed by the pre-compiler, and affect ALL other files processed after
them.

This is completely wrong. It's not even a matter for debate. Learn C before you comment on it.

But even if you were right, the solution is obvious to anyone who understands C: Scope the macros with an #undef.

One of the worst macros ever that I've had to deal with was in the
Mersenne Twister code written by Takuji Nishimura and Makoto Matsumoto
(http://www.math.sci.hiroshima-u.ac.jp/~m-mat/MT/MT2002/CODES/mt19937ar.c)
that a programmer put into our code-base and subsequently had gobs of
error messages out of [seemingly] "nowhere". These lines:

#define N 624
#define M 397

Caused no end of headaches for many hours. Your macros are of a similar
vein. And since you can't guarantee compile order, you are just asking
for problems. What if another programmer later decided to do this:

void some_func(int _x) {
  /* do something here */
}

My macros are *nothing* like the example you provided. The names I picked have a simple and obvious relationship to the structure members. The macros you have don't have any obvious relationship to anything.

But ignoring that, I have no idea what your "guarantee compile order" comment means, because this is a single file and I *can* guarantee compile order. As far as what prevents someone from writing the function you offered, it's called a compiler and the mechanism is a syntax error. The expansion of your function would be this:

	void some_func(int this->x) {
	  /* do something here */
	}

And that would immediately fail as a syntax error. At that point, the clueless programmer would scratch their head and actually *read* the code and understand the conventions being used.

The *only* sin here is not adding a comment explaining the rationale for the macros. Since my purpose was to illustrate that adapt2.c could be factored further, and since this wasn't intended to be production code, I didn't care.

This is now an error because of your macro. But obviously your
standards are different. Okay. ;)

It's more than an error-- it's a *syntax* error that the compiler wouldn't even compile.

Are you seriously suggesting that "good programming" is the ability to write functions without understanding the environment they are compiled in? Man, you must really hate Forth then, as you're free to change the environment in all sorts of interesting ways-- far beyond mere macros in C.

Of course, the debugger I have is fine. I certainly *could* go to the
top of each of the functions that is called, set a breakpoint, and set
one at the end. What a pain! Also, that actually requires me to have
access to the source code of those functions. I very well may be using
a 3rd party library, in which case your argument is moot.

What a pain?!!! Do you really understand how to use your debugger? Do you understand that when you put a breakpoint on a line, what you are *really* doing is identifying to the debugger a displacement into your compiled code and setting a breakpoint there?

Your argument is nonsense on multiple levels. But if it really bothers you and your debugger is insane and strictly line-based, a moments editing solves the problem:

	while (!
	    converged(
	        newPoints(
	            moveDown(
	                subdivide(&i)))))
	;

Your comment about not having source code and third-party functions also doesn't make sense. We aren't talking about cases where you don't have source code-- we're talking about debugging *this* code.

Now, how much easier is it for me to just set one breakpoint at the
first part of my code that begins using those functions and gets the
results of each and just passes them on? There is little [to no]
performance hit, and only makes life easier for anyone maintaining the
code in the future. [...]

Sounds like you're someone who doesn't operate below the C level and you rely on a debugger operating at a high level of abstraction to insulate you from the machine-level reality underneath. Your need to reformat source code to set breakpoints is a clear sign of that.

So I guess in a weird way, you've brought up a valid point; if I am dealing with programmers who can't operate below the level of C, then you're right-- reformatting the source is necessary. Thankfully, my reformatted version above addresses that, and even more thankfully, being an embedded systems programmer, I don't really have to deal with that issue at all.

Readability != maintainability. They may be closely related (in good
code), but they are not the same. And when working on a project that
takes many years, with programmers switching between systems/projects
and new hires, maintainability is far more important (imo). YMMV.

I didn't claim readability is the same as maintainability. But I do believe that if your code isn't readable, then it isn't (by definition) maintainable.
.

Relevant Pages

  • Re: Include Statement
    ... For the same reason conditional compilation directives were added to compile a block vs not to compile a block. ... Second, it is difficult to explain all the reason because they are VAST, every programmer has their reasons. ... Like without out product line source codes, look at other large projects where the source code is LOADED with both imports and includes - because the language allowed it. ...
    (microsoft.public.dotnet.languages.vb)
  • Re: Schadenfreude by proxy
    ... and where would you find the source code for one should I wish to ... compile my own? ... when a boy programmer and a girl programmer love each other very ... If that's the metaphor for open source, then what's the metaphor for closed ...
    (alt.sysadmin.recovery)
  • Re: Schadenfreude by proxy
    ... and where would you find the source code for one should I wish to ... compile my own? ... when a boy programmer and a girl programmer love each other very ... If that's the metaphor for open source, then what's the metaphor for closed ...
    (alt.sysadmin.recovery)
  • Mystified by compiling
    ... I'm not a programmer so pardon me for this query but I have ... this source code that I want to compile as a MS-DOS ... randgen.cpp:1166: error: 'getcwd' was not declared in this scope ...
    (comp.os.msdos.djgpp)
  • Re: C++/TCL Need Solution to Compile Error c2784
    ... There's always the "Platform SDK" from MSDN if you need the window.h stuff, ... Since I'm using the Command Prompt rather than the GUI to compile the ... I then tried to compile your source code and I kept getting the same ...
    (comp.lang.tcl)