Is it a lost cause?

That package of RTB was made before the newer /dev/gpiomem interface was added into the kernel and you're right - it's not the best thing to do, however RTB drops root privs. immediately after initialising wiringPi which is essentially the first thing it does in main() after getopt.

It needs to do the getopt thing to work out what way to initialise wiringPi - BCM pins, board pins or wiringPi pins.

After that it's running as the user that called it, so things like file access, creation, etc. work as the calling user.

Check this by running it, then typing in:

x = openOut ("/tmp/testfile1")

then (in another terminal)

ls -l /tmp/testfile

and checking the ownership of the file created.

I'm doing some work on it so will bear that in-mind for the next time I make a package.

Alternatively, if you remove the set uid bit,

sudo chmod 755 /usr/bin/rtb

it will still work, just not initialise wiringPi for GPIO access.

One thing you can't do with /dev/gpiomem is change the internal pull up/down resistors, so if you need that, you need root.

Gordon (Author of wiringPi)

Your (desktop?) Linux system might do that today, but the Pi didn't 4 years ago.

Gordon

I Expected that to be the case.

Indeed my comments were meant to be taken constructively.

Can't help, I'm afraid since I've never tried to extract sound from my RPi.

Yes, that sounds like something I might also do in C or Java.

Interestingly, I can read the first example quickly; not the second. I spend more time matching lines in the second example. I guess readability depends on experience.

/BAH

You have just found the reason that standards manuals limit structured programs to having subroutines no more than 60 lines (=1 page) long whereas GOTO programs can be 20 pages long.

My 'END' statements had a comment saying what it was attached to.

That limit would be a problem in my work world :-) I never had problems reading code which spanned more than one page. My programming style would use dispatch tables when entering a module for the first time.

that's nice if it's kept up to date. In my area, it would be out of date with the third programmer. We would put a stand-alone comment at the end of a logical piece of code to indicate the last page of the routine.

After each monitor edit, one of the procedures was to run a TECO macro over the sources to match angle brackets.

/BAH

I do this too. Often it's a pain in the neck, but occasionally it's useful. IBM PL/I has a nice "nest" compiler option that prints the containing block and group number for each statement, so you can easily visualize the structure.

Often it is easier, though less efficient*, to break deep nesting into subroutine calls, so each level of logic has its own 'block'..

That happens to me, particularly if the code contains a 'switch' statement and contains other stuff such as another 'switch', or conditionals.

That is my solution.

Which is what I do.

The nice thing about modern editors is that they'll automatically associate end with the appropriate begin with a simple keystroke and/or highlighting (e.g. with the enhanced vi (vim), when the cursor is over a begin (e.g. { for C), it will highlight the corresponding end (}) and the '%' key will move the cursor between them.

Burroughs BPL language had 'END WHILE', 'END CASE', 'END IF' etc. in the language.

"A compiler", perhaps. "The compiler", well you need to be more specific.

In C, a static function has file scope, so it is not eligable to be a target for a call from another compilation unit, so there is no impediment to generating it inline. Although, with gcc, this should only happen if compile time optimization (-O) is requested.

If(brexit) { } //endif (brexit) etc etc.

I stick em in the comments otherwise I haven't a clue;)

Rather a sweeping statement. It will depend on the compiler. One might say that logically the compiler should always do this, but I can think of situations when I would rather it did not.

I'll leave identification of such a situation as an exercise.

A few decades ago, MAINSAIL (MAchine INdependent Stanford Artificial Intelligence Language) allowed an optional string constant after BEGIN and END. If used, the compiler would enforce that the string constants match. That was quite helpful on occasion.

Alas, no. At least not in the newest full Raspbian from the download page, dist-upgraded.

"2016-05-27"

It used to be there... Oh well.

Gordon

If you got to where you are by upgrading from wheezy to jessie rather than reformatting the SD card and installing jessie then you won't have rtb - that's what I did and it wasn't on my RPi when I looked, but running "apt-get install rtb" added it to my system.

However, unless Gordon has refreshed the rtb .deb package recently the example programs and manual will not be where the manpage says they should be. I'm unsure where to find the example code but the manual is a PDF linked from the bottom of the Return To Basics history page on