View previous topic :: View next topic |
Author |
Message |
kory Guest
|
Anyone tried 3.202? Seems to be really broken |
Posted: Wed Jun 02, 2004 12:06 pm |
|
|
I tried 3.202 lathe night at it is worse than 3.201. Genereates buggier code. Unfortunately, 3.201 also has one bug that has no work around for me, so I'm totally stuck right now. 3.191 had two bugs.
Kory |
|
|
Guest
|
Re: Anyone tried 3.202? Seems to be really broken |
Posted: Wed Jun 02, 2004 1:45 pm |
|
|
kory wrote: | I tried 3.202 lathe night at it is worse than 3.201. Genereates buggier code. Unfortunately, 3.201 also has one bug that has no work around for me, so I'm totally stuck right now. 3.191 had two bugs.
Kory |
If you post a comment like this could you please back it up with examples.
Just what is broken in 3.202 and What are these special bugs that stop you using 3.191 ?
Tried it today and it compiled 48K (of 64K) good code... |
|
|
kory Guest
|
Re: Anyone tried 3.202? Seems to be really broken |
Posted: Wed Jun 02, 2004 2:18 pm |
|
|
When I posted the first message, I had not debugged the compiler code yet. What I've found so far is that in a large switch()/case statement, the compiler generates an absilute jump into a completely different function and causes the chip to reset.
One other person on the forum had the same prbolem and the #opt 5 pragma solved their problem. I tried it and it make no different for me (the code this get bigger). I even tried lower levels of 4 and 3.
I hit this problem pretty early on, so I have not been able to exercise much of the code.
Kory |
|
|
William H. Conley III
Joined: 27 May 2004 Posts: 17 Location: Tucson, AZ
|
|
Posted: Wed Jun 02, 2004 2:22 pm |
|
|
"Kory"
You don't have to post your code on the forum but if you could email me what you have I maybe able to help. |
|
|
C-H Wu Guest
|
Re: Anyone tried 3.202? Seems to be really broken |
Posted: Wed Jun 02, 2004 8:04 pm |
|
|
Kory:
kory wrote: | I tried 3.202 lathe night at it is worse than 3.201. Genereates buggier code. Unfortunately, 3.201 also has one bug that has no work around for me, so I'm totally stuck right now. 3.191 had two bugs.
Kory |
Would you like to try 3.187 with #opt 10 ? It works pretty good for me.
3.187 does not have the bugs found in 3.188 ~ 3.201. The only potential bug for 3.187 worries me is "Problems with very large PIC18 data structures (fixed in 3.189)"
Best wishes
C-H Wu |
|
|
Guest
|
Re: Anyone tried 3.202? Seems to be really broken |
Posted: Fri Jun 04, 2004 3:54 pm |
|
|
kory wrote: | What I've found so far is that in a large switch()/case statement, the compiler generates an absilute jump into a completely different function and causes the chip to reset.
Kory |
Just how big is the switch ?
The code I'm working on at the moment has switch with 37 cases and it working just fine.
Hans W |
|
|
Guest
|
Re: Anyone tried 3.202? Seems to be really broken |
Posted: Fri Jun 04, 2004 7:44 pm |
|
|
Anonymous wrote: | kory wrote: | What I've found so far is that in a large switch()/case statement, the compiler generates an absilute jump into a completely different function and causes the chip to reset.
Kory |
Just how big is the switch ?
The code I'm working on at the moment has switch with 37 cases and it working just fine.
Hans W |
It is fixed
Kory wrote: | Darren's suggestion of using an int16 in the switch statement cleared the problem! I even removed the #opt 5 and it still works! Thank you, Darren!
I still have the annoying problem in that I can't load non-debug code with the ICD-40U, but for the work around, I use the ICD2 to burn the production boards.
Thanks everyone for your help. |
see the end of this: http://www.ccsinfo.com/forum/viewtopic.php?t=19488&postdays=0&postorder=asc&start=15&sid=e1d79e35ecb7ddb79a4edc48a815c777 |
|
|
|