Version 4 Comments

[olivercox]

I've been leaving the new "improved" [sic] interface hoping that it might improve - no such luck!

I've also just renewed my maintenance agreement but am beginning to wonder what I'm paying for!

CCS has been such a wonderful compiler for so many years it is really sad to see what is happening yet again with a new version.

Please please please CCS stick to what you're good at, put your effort into making the compilers work well and forget the IDE. Just make the old IDE work with the new compilers properly for those that want to use your IDE, and leave the rest to use the MPLAB IDE (which is improving all the time anyway).

[MikeW]

I am also getting very fed up with CCS


I need to work with a PIC16F886.

the V3 wont work with it, the V4 020 is still poor.

i am close to giving up, the Mikroelektronic is becoming more interesting ...


I rennewed my maintainence agreement just to get 886 compiler, thus far, it aint worth it

[Martin Berriman]

[ckielstra]

[Martin Berriman]

New version out today 4.021 however it still has issues (see link)

[arunb]

Hi,

Is Microchip's version better than CCS ???

thanks
arunb

[Martin Berriman]

[arunb]

Hi,

I have read the link, I think the PCH version should be OK for most of my projects, but after reading thiese posts I am having doubts, are the other versions (apart from 4) any better ??

thanks
arunb

[Ttelmah]

Yes.
Basically, 3.249, 99% works. It has some faults (for instance, the support for the 'J' chips), and lacks some of the new features (though some are present, and generally work better than in the V4 releases so far. Most users here writting code that they want to actually work, are using this version.
The core problem is that CCS, have an attitude to 'new releases', that should be a capital offence (possibly downgradeable to just skinning alive...). They appear to have so real setup to beta test anything. What they put up as 'new releases', are at best 'betas'. At the start of V4, they launched what they claimed was a 'beta test' program, but versions offered were poor alpha releases, and the product was marketted, before even 1/4 of the bugs were fixed. Currrently the V4 releases are starting to work for older code, but as yet few of the 'new features' produce good results. Even worse, support for V3, was dropped months before the launch of V4, and the known bugs, though often only minor, remain unfixed.
I'd suspect the core problem is actually bad documentation, and a poor idea of program structure. I actually wonder if anyone at CCS, has even heard of a compiler 'T' diagram....

Best Wishes

[anauta]

Hi,

After some trouble with 4.021 (verification errors) I just tried the latest 4.022.

No flash verification errors anymore

However, it fails to write to EEPROM properly (strage or no write at all).

PIC4520, Windows XP SP2, PCWH 4.022.
Just went back to 4.020, no errors, so I am happy...

Greetings,
Auke Nauta

[anauta]

Hi,

Latest versions will not correctly write float's to eeprom:
Version 4.020 is OK.

Anyone else has the same experience?

Greetings,
Auke Nauta

Code:
------
float read_eeprom_float(int adr){
int *k,i;
float f;
for (i=0;i<4;i++){
k=&f+i;
*k=read_eeprom(adr+i);
}
return f;
}

void write_eeprom_float(int adr, float f){
int *k,i;
if (read_eeprom_float(adr)==f) return; // no change, exit
for (i=0;i<4;i++){
k=&f+i;
write_eeprom(adr+i,*k);
}
}

[Ttelmah]

Your code would not work on any compiler other than CCS, so it might well be worth trying 'proper' C syntax, to see if V4, is actually behaving _properly_ at last wth regards to pointers!...
The 'point' is that when you increment a pointer in C, it is _meant_ to move to the next object of the type 'pointed to'. To move through the bytes, then requires you to cast the pointer to a 'byte', 'int8', or 'char' pointer, before the increment. So:

[anauta]

Thanks ttelmah,

Yes, the code works.
Apparently CCS changed this behaviour from version 4.021.

Greetings,
Auke

[Ttelmah]

Very interesting.
Many moons ago, CCS switched to the 'proper' syntax, on one version of V3. It caused so many things to stop working (a lot of their 'demos', use their own incorrect syntax), that they quietly changed back a couple of versions latter. This is why, whenever I post code using pointers, I always explicitly force them to address an 'int', and do the arithmetic myself, since this way it'll work whichever way the compiler behaves.
Now, if CCS are getting this completely 'right' now, the behaviour should switch back, if you select 'CCS3' for the compiler syntax. If so, then this part of the new compiler really is orking just as it should.

Best Wishes

[Darren Rook]