Making PBP code more modular

[forgie]

Hi there,
I've been using PBP on and off for a few years, now semi-professionally. I'm interested in what coding techniques people use to keep their PBP code modular and portable (i.e. the way that you code using C for a PC)...... or should I be using a C compiler to satisfy my desire for modular code? I find that things become unwieldy and messy when I start splitting projects into multiple files, as far as variable/pin declarations go.

I don't want to start the tired old BASIC vs C flamewar here, I just want to know if anyone else uses multiple files for their PBP projects in a way that is properly modular/portable as well as being very readable.

Any code examples/tips/tricks/whatever would be appreciated.

Thanks,
Nick Forge

[mister_e]

Personnaly, i never use Include in my code, i prefer to see everything in my face. I have several routine that i built and i just copy/paste them in my code.

[NavMicroSystems]

I agree with Steve,

I have a number of SUBs I use on many projects.

When I started I used to include them.
But today, I cut and paste them into my projects for better readability.
That way I have everything at a glance.
(No need to look at different files)

It is no real problem to write modular code in PBP.
agreed, VARs are global,
but with some creativity you can write modular and readable code
and minimize RAM usage.

[forgie]

I just feel much more comfortable having 6 files of around 50-100 lines each, each file having a specific task to complete, rather then having one monolithic 400-500 line file. Maybe I just have to break out of the PC programming (C) mentality, where you try and hide code away somewhere else (inside a module) once it is finished and debugged - this is perhaps one of the reasons for the "PBP /= Professional quality" attitude that many have: on a PC one always strives for modular, portable code 100% of the time. On a PIC, by the very nature of the beast that isn't always possible.

Do you guys follow PC convention in so far as removing 'magic numbers' from your code and using constants to control registers? By this I mean saying something like:

SLEEP_PORTE CON %00000101
WAKE_PORTE CON %00000010
~

Sleep:
PORTE = SLEEP_PORTE
Sleep
PORTE = WAKE_PORTE
Return

This gives you a convenient place up the top where you can define the PORT states for each 'program state'. Just to give an idea of where I'm coming from.... maybe I need to completely reset my perspective on this whole issue. (Could someone please pull the MCLR attached to my mentality to ground?)

[Melanie]

I've written many programs, the biggest almost six and a half thousand lines (that's a 200kb .BAS file which reated a 8Mb .LST by MPASM!!!)... Like the other replies, I never use include files, but my programs stick to a very rigid and repeatable format... so it is 'modularised' in that form...

1. Define Configuration Fuses
2. Define Hardware (usually sequential in Port order)
3. Define EEPROM (in address order)
4. Define RAM (always in alphabetical order)
5. Define Constants (always in alphabetical order)
6. Subroutines (always in alphabetical order)
7. Hardware Initialisation
8. Main Program Loop
9. Interrupt Handlers (always in alphabetical order)
10. Error Handlers (always in alphabetical order)

I don't redefine things that are already included by PBP... ie things like TRISB or OPTION_REG, but I will use the same Label Names as are in the Datasheet and am consistant to use the same variable, hardware or constant names across all my programs if they are doing the same job. Once I have a routine for a particular function (ie reading a keypad or ADC), that routine is replicated EXACTLY and I don't reinvent the wheel. Changes like Timeout values are always predefined constants. Section 2 in my program structure (Hardware Definition) will reassign the pins so that my Keypad routine for example will still work without change or worry regardless of the PIC or PCB layout.

Keeping to a consistant structure allows me to create programs quickly, with the minimum or errors, allows me find what I'm looking for and best of all, allows me to refamiliarise myself with the program easily, when it needs to be revisted several months down the road. Life is too short, if you have a formula that works, flog it for all that it's worth... (here we go, back to whips and things again)...

[cupajoe]

I write a bit of software and I very seldom use includes. I follow a routine similar to Melanie's in that I tend to structure all of my programs the same way, pretty much as Melanie described. One differance, I usually write my ISR in assembly and stick it at the very top. I also define magic numbers (constants) for things like timer preloads and initial variable states. I also use aliases to define registers. This way my code looks the same, regardless of PIC, because I try and use the same names. I too find that I cut and paste routines when writing code.

All in all code portablity is pretty good but remember we are talking about micros. I mean, its not like you are gonna run your PBP code on an AVR! Now that would be portable!

Regards,

Joe.

[forgie]

Following the PBP experts lead I went through the task of converting my 7 files project into a single, big momma file. In the process it has helped me tidy up alot of my PBP coding style as well as variable naming. It's taken me a while to adjust to a different style of indenting (and the visual appearance of PBP code is so different to C style code it's really hard to find it readable without doing a hell of a lot of it):

If oranges = apples Then
Do something
Endif

vs

If (oranges = apples) {
Do something;
}

Overall I am much more happy with PBP as a language since forcing myself to do it the 'hard' way. I can see perfectly well why C-experienced PC-programming people don't take to it, but I think it's perfectly up to the task that it has in front of it (programming pics, not multimedia software packages).

Thanks for the tips guys,
Nick

EDIT: damn these auto-formatting forums.... the guys who wrote the code must have thought they were being so clever. oh well i hope you get the picture of what i tried (and failed) to show. if not.... well it doesn't really matter, does it?

[Darrel Taylor]

WOW, this is very surprising. Nobody uses Includes???

I can't imagine NOT using them.

Reasons for my thinking. :

-- Version Control -- Back when I wasn't using Includes, I'd have the same routine Cut&Pasted into several different programs. While working on one of them I'd have a brainstorm and figure out a better way to do it. Now all the other programs are either stuck with the old version, or I'd have to go back through every program I've written in the past to update their in-line code. Ugg.

-- Thought Grouping -- As in Melanie's modularized list, Typically any variables, aliases etc. will be placed at the top of the program. But more than likely, the routines in question will be buried deep in the middle of the program, completely isolated from all the other things that go with it. By using Include files, you can keep things grouped together so that you always know what variables or other subroutines are needed to be able to use the said routine.

-- Ease of use -- If I need an interrupt driven clock or a buffered UART routine. All I need to do is type in 1 or 2 lines. No cut&paste, no variable searching, nothing. Just boom, there it is.

I always wondered when I wrote an include file for sombody else, why they would take it apart and paste it in 4 different places in their own program. Just doesn't make sense to me. And seemed rude to me at the time.
But apparently, everybody would do the same thing.

----------------

And, about those "Magic Numbers", don't like 'em
Do you know how many problems on both the List and the forum have been tracked down to Magic Numbers. The newbie just cut&pastes the code we give them without having a clue how it works. They spend a week trying to figure out why the Code doesn't work, and it turns out to be that T1CON=%00001001 should have been T1CON=%00000001 or something similar.

Granted, doing it all in one shot saves instruction cycles, and code space. But it sure can cause problems when sharing your code in public forums.

Best regards,
   Darrel Taylor

P.S. scroll down to the bottom of the page and click on the vB Code link. There are several ways to format your posts.

[cupajoe]

Darrel,

I guess that I just never got into the habit of using includes. I almost always have an assembly language ISR at the begining of my programs and I could probably make different ones up with different features and then just include them. But I guess I just never took the time to make sure it would work properly so I don't trust it. I might try it on my next program. If I do I will post a follow up.

As for magic numbers...I am not sure you follow. A magic number is a nicely named constant that allows you to more easily control your code parameters from one place. Magic numbers make your code easier for newbies to understand because the numbers are given meaningful names.

Your example of
T1CON=%00000001

might look like this
'CONSTANTS and MAGIC NUMBERS
TMR1_ON = %00000001 'constant to plug into T1CON to start timer
TMR1_OFF = %00000000 'constant to plug into T1CON to stop timer
Some really nice code goes here!
Main:
T1CON = TMR_ON 'start timer1
Some more code.
T1CON = TMR_OFF 'stop timer1
goto main

I usually don't do it this way. I usually alias the register bit and set it equal to 0 or 1.

TMR1_enable = T1CON.0 'set TMR1_enable to start clr to stop.
TMR1_enable = 1 'start timer 1
TMR1_enable = 0 'stop timer 1

A better example of magic numbers would be
'Constants--------------------
buffer_size = 16 'This is how big the buffer is (in chars).
'Variables----------------------
buffer var byte[buffer_size] 'This is were the serial in characters go.


I think that these examples show that magic numbers can make it easier to make changes to your code and they make your code easier to read.

What do you think?

Regards,

Joe.