Version 1.11.7.4 - comments after 15 years of c++ experience

ALJO · Joined: 17 Feb 2005 Posts: 2

I have over 15 years of experience in c/c++, assembler and some basic. 7 years out of those 15, I have been developing 3D games for PC.

With my new job, I had to test the BASCOM 1.11.7.4 for the ATMEGA 128 (to see if it would work) and I would like to share some interesting discoveries about the version 1.11.7.4.

The BASCOM compiler compiles the code depending on the weather and the current concetration of moisture in the air. Let me explain.

For example. Variables passed through the function, were not always there. Sometimes they are, sometimes they are not. It feels like as the BASE POINTER would get lost. Those who know assembler, would have an idea what I talk about. Ok never the less... Let's say i was drunk. It's my fault.

Then. Interrupt #3, does not work on falling edge (like all others), but only on LOW LEVEL. Hmm... Let's say it's my fault again for some reason. But, how would you explain the fact that all other interrupts worked ok, on any settings FALLING, RISING, LOW LEVEL, execpt Interrupt #3 on LOW LEVEL. It disregarded the setting for falling edge. As if I haven't written the code. It ignored the statement totaly.

Ok, let's say is still my fault again. But, then what about this one. One global variable, named C3_ready (that's how I called it). Was seen in two different ways.

In the main loop it was 0 (zero) in the interrupt routine it behaved as 1 (one). And this lasted constantly weather interrutps were happneing or not.

Do not understand, that this was a split second... No no... This insane situation stayed like that permanently with no further interrupts occuring at all.

Now, someone would think I mixed two variables names... Perhaps I had an array going out of its boundaries, overflow, or that I HAVE NO IDEA HOW TO WORK MULTI THREADED program, and that the time-overlaping was playing with me on the variable values. Yeah right... Have been living with multithreaded programs for last 10 years. I work up to 10 threads at the time.

I have written programs 10.000 times larger in my life... and I eventually manage to remove any such bug. But here, I had only few pages of code and again the insanity of BASCOM 1.11.7.4.

Even more shocking than that. I have changed the variable NAME (!!!) from C3_ready to X3_ready and the program worked ok. Then I changed it back to C3_ready and the program had the same problem. For the same variable it saw 2 different values again (zero and one). Insane !!

And this variable appeared only in 4 different locations in the whole program. I also had another two similar variables C1_ready and C2_ready. Those two worked ok (with interrupts !!). But, C3_ready DID NOT! Bascom did not like the name. It had to be called X3_ready, instead of C3_ready and then it worked fine.

I got absolutely blown off my mind. I have no comments.

I have got more of such discoveries, but have no will to talk anymore... Does anyone know who wrote this evil thing called BASCOM ?

AL-JO

Frankeman · Joined: 11 Aug 2004 Posts: 948 Location: the Netherlands

Hi,

Welcome to this friendly forum !

After 15 years of experience in programming you know that you must use the latest software version.
At the moment Bascom is at version 1.11.7.7

I do not know if the problem you experience is fixed in this version, but if the developer of Bascom reads your story about a hard day at the office he is willing to repair it.
(i never met him but i think he is human)

Frank.

Luciano · Joined: 29 Nov 2004 Posts: 3149 Location: Italy

Try this product:

http://www.iar.com/FilesPublic/EW/002114/DS-EWAVR-410.pdf

Best regards,

Luciano

Evert :-) · Joined: 18 Feb 2005 Posts: 2156

After 15 years experience of reading manuals;

- You can add _ at the end of a line to indicate that the line will be continued on the next line.
For example : print "this is a test" _
print " this is part of the previous line"

So, C3_ready is not a valid variable.

Regards, Evert

P.S. Try http://www.imagecraft.com for a C compiler

Juergen · Joined: 13 Nov 2004 Posts: 85 Location: AZ, USA

did you write programs at Microsoft? Sounds like it with all the newbe errors you made.... Innovation replaced by FUD.

We have several commercial programs written with BASCOM and never had such problems, neither in 1.11.7.4 nor 1.11.7.7 (maybe because my programmers are able to read manuals before they attempt to write C in a BASIC compiler?).

Juergen

philm · Joined: 14 Jul 2004 Posts: 138 Location: Australia

I have over 40 years of experience of living. One thing that I have learnt is that badmouthing and whinging like a little girl will get you nowhere.

Werner · Joined: 27 Sep 2004 Posts: 30 Location: Germany

Hallo Evert,

ALJO · Joined: 17 Feb 2005 Posts: 2

Thanks for all your replies. Talking about the C3_ready not being a valid variable name... Can you answer then, why:

C1_Ready was a valid name and it worked fine ?
C2_Ready was a valid name and it worked fine ?
X3_Ready was a valid name and it worked fine ?
But, C3_Ready was not ?

Still my fault ?
Al-jo

Luciano · Joined: 29 Nov 2004 Posts: 3149 Location: Italy

In VB the line-continuation character is a single leading space followed by an underscore ( _).

Luciano

Luciano · Joined: 29 Nov 2004 Posts: 3149 Location: Italy

More test.

(The second example is just to show the limits).

Regards

Luciano

'=========================
'Example 1
'OK with VB 6.0
'OK with Bascom 1.11.7.4

Dim My_variable_a As Integer
Dim My_variable_b As Integer
Dim My_variable_c As Integer

My_variable_a = 1
My_variable_b = 2
My_variable_c = 99

My_variable_c = My_variable_a + _
My_variable_b

Print My_variable_c

'=========================

'=========================
'Example 2
'OK with VB 6.0
'NOT OK with Bascom 1.11.7.4

Dim My_variable_a_ As Integer
Dim My_variable_b_ As Integer
Dim My_variable_c_ As Integer

My_variable_a_ = 1
My_variable_b_ = 2
My_variable_c_ = 99

My_variable_c_ = My_variable_a_ + _
My_variable_b_

Print My_variable_c_

'=========================

Calv · Joined: 08 Dec 2004 Posts: 26

I am considering a "I love BASCOM" sticker at my car.

Calv · Joined: 08 Dec 2004 Posts: 26

I am considering a "I love BASCOM" sticker at my car.

DToolan · Posted: Fri Feb 18, 2005 8:26 pm Post subject:

exidir · Joined: 22 Feb 2005 Posts: 1 Location: Australia

Hi All,
talking about the bugs of a compiler or any kind of software has nothing to do with whinging. It is more about finding the reasons for unexplainable behaviour and maybe get the developper to have a closer look at it.
We have used Bascom for a number of quick projects and have found similar hick-ups as Aljo. I know there are a lot of heros out here, that have never found any problems in their software, but maybe they have also never done a full test of their programs. We had variables that were simply ignored by Bascom. When we disassembled the code and ran it in Atmel Studio, we found that the compiler had virtually ignored whole modules of the code.
It took me 2 days to work out, that Bascom simply does not do If-then-elseif statements, even though it claims to do it. In the end I copied the sample code and dissassembled it. Again, the whole bit of the elseif statement went missing in the compiler.
We do still use Bascom and are normally quite happy with it for small projects, but every now and then you find some real hick- ups in it.
Bascom for production code certainly requires a lot of intensive testing before the release, but that is also true for avr-gcc and similar compilers.

Has anyone ever successfully used inline assembly code in the way it is described in the manual?? Give it a try, it leads to some very funny results.
All this is based on my experience with Bascom 1.11.7.4 which was the latest version at the time of the project.
Exidir

Ingo · Joined: 28 Jan 2005 Posts: 46 Location: Germany

Hi all,

whining does not lead anywhere, also accepting errors as a fate is no solution. Errors should be named clearly and free of emotions together with the circumstances under which they appear. Then, and only then, the programmers have a chance to reproduce and to fix them. Compilers are complex things in which it is not possible to test everything under any possible condition. Proper feedback from the filed is needed for continuous improvement.
Even high professional and much more expensive compilers have bugs that are treated this way.

Hoping for further improvements,

Ingo

PS: *fast copy* and AVRco have bugs, too!