First F8 experience

[Jan Lichtenbelt]

Flowcode 8 looks very nice. Thanks to all who made this possible!

My first flowcode 8 project gave some warnings errors like:

C:\ProgramData\MatrixTSL\FlowcodeV8\CAL\PIC\PIC_CAL_I2C.c:
212: #define PIR3 PIR4
^ (111) redefining preprocessor macro "PIR3" (C:\PROGRA~2\Flowcode\Common\COMPIL~1\pic\include\pic16f1847.h: 663) (warning)
Accu_Activator_V49_F8_GLCD.c:

using updated 32-bit floating-point libraries; improved accuracy might increase code size
C:\ProgramData\MatrixTSL\FlowcodeV8\CAL\PIC\PIC_CAL_ICD2.c: 188: non-reentrant function "_MARKER" appears in multiple call graphs and has been duplicated by the compiler
idem "_PUSH" ,"_POP", etc

0: (1347) can't find 0x22C words (0x22c withtotal) for psect "text20" in class "CODE" (largest unused contiguous range 0x197)
updated 32-bit floating-point routines might trigger "can't find space" messages appearing after updating to this release; consider using the smaller 24-bit floating-point types
0: (1347) can't find 0x1FC words (0x1fc withtotal) for psect "text18" in class "CODE" (largest unused contiguous range 0x197)
idem some other "textNN"


0: too many errors (31)

Who can help?

With kind regards

Jan Lichtenbelt

[Benj]

Hi Jan,

Can you post your program so we can investigate. Sounds like your out of RAM or ROM on the device?

[Jan Lichtenbelt]

Hi Ben,

Please find attached the file. It works fine in Flowde 7.


Kind regards

Jan

Accu_Activator_V49_F8_GLCD.fcfx

(95.14 KiB) Downloaded 290 times

[Benj]

Hi Jan,

It's compiling ok for me in v8 until I enable ICD and then the code is too large to compile due to the extra code added to allow ICD to function.

212: #define PIR3 PIR4
^ (111) redefining preprocessor macro "PIR3" (C:\PROGRA~2\Flowcode\Common\COMPIL~1\pic\include\pic16f1847.h: 663) (warning)

This is slightly worrying but it's from the Microchip XC8 compiler header file so it could be ok or it could be a bug in the compiler?

Edit, no sorry it's not the XC8 compiler it's our CAL file, I've added a fix for this now and will add it to the updates system shortly.

[Jan Lichtenbelt]

Hi Ben

After an update of Flocode 8, the first error (PIR3) has been solved. But now the second error is still there

0: (1347) can't find 0x18 words (0x18 withtotal) for psect "text49" in class "CODE" (largest unused contiguous range 0xD)
updated 32-bit floating-point routines might trigger "can't find space" messages appearing after updating to this release; consider using the smaller 24-bit floating-point types
0: (1347) can't find 0x16 words (0x16 withtotal) for psect "stringtext2" in class "STRCODE" (largest unused contiguous range 0xD)
0: (1347) can't find 0x16 words (0x16 withtotal) for psect "stringtext3" in class "STRCODE" (largest unused contiguous range 0xD)
0: (1347) can't find 0x14 words (0x14 withtotal) for psect "swtext1" in class "CONST" (largest unused contiguous range 0xD)
etc etc

Can you help with this error?

Kind regards

Jan

[Benj]

Hi Jan,

These errors are because you have ran out of memory on the target device, hence the program cannot fit.

Can you select a device with more available memory?

Did the same project on the same target device compile on v7?

I've just tried your program and it's compiling correctly for me in v7 and v8. v8 on the left in the screenshot.

CompileStats.png

(16.66 KiB) Downloaded 2574 times

[Jan Lichtenbelt]

Hi Ben,

That is curious! I have still the same problem for FC8. F7C works fine. Please find the mgs.txt file FC8 attached.

Kind regards

Jan

Accu_Activator_V49_F8_GLCD.msg.txt

(6.38 KiB) Downloaded 269 times

[Jan Lichtenbelt]

Hi Ben

I tried this program also on my Laptop, but unfortunately with the same error message.

Information about my PC and Laptop is:

PC: Windows 10 Pro
Version 10.0.17134 build 17134
x64-bades PC
Pentium Dual Core processor E5300 2,60 GHz

Laptop: Windows 10 Home
Version 10.0.15063 buikld 15063
x64-bades PC
Intel processor T5600 1.83 HGz dual core

I hope you can find a solution.

Kind regards

Jan

[kersing]

I've just tried your program and it's compiling correctly for me in v7 and v8. v8 on the left in the screenshot.

Ben,

It looks like you are (unintentionally) cheating. The FC8 size matched perfectly once I modified the compiler settings to use 24 bit float. FC8 uses 32 bit float by default.

Best regards,

Jac

[kersing]

Jan,

Place the attached batch file in c:\Program Files(x86)\Flowcode\Common\Compilers\pic\batch and in the Menu select 'Build->Compiler Options'. Modify the 'Compiler Location' as shown below, hit the save (disk) icon and specify a unique name. That should allow you to build your code.

image00096.png

(14.56 KiB) Downloaded 2480 times

[QMESAR]

It looks like you are (unintentionally) cheating. The FC8 size matched perfectly once I modified the compiler settings to use 24 bit float. FC8 uses 32 bit float by default.

Thanks Kesing was not aware that the 32bit is the defualt in FC8

Team as you are aware that the ICD has a lot of issue on PIC16 and 18 with the 24bit float in FC 7 you actually made a post how to go to use 32bit float in XC8 to fix the issue we were seeing on the ICD

Here http://www.matrixtsl.com/mmforums/viewt ... 68&t=19700
I will throw my toys out of the window if you change things and we are back with numbers of ICD issues

[Benj]

Hi Jan,

I tried this program also on my Laptop, but unfortunately with the same error message.
I hope you can find a solution.

It's compiling for me ok in v8, however if I enable ICD then this no longer compiles as we have run out of ROM. ICD requires the use of ROM to add in the functionality to communicate to the ICD host and so we know where in the program we are. So the program is ok without ICD but is using too much memory (90%) to allow ICD to be switched on. I don't think this is a bug it's kind of to be expected. Maybe there is a larger memory device you could use temporarily to do the ICD testing and check your code is ok?

Hi Kersing,

Thanks for the alt batch file, I'm sure some users will require this to save on some memory usage, but really the savings are minor, I'm not too sure why Microchip bother with their weird and wonderful variable formats as it just makes the software none standard and horrendous to integrate into other things. Maybe it's to try and lock you in to using PICs? More likely it's to squeeze out a few more instruction cycles.

Hi QMESAR,

Currently no plan to go backwards and 32-bit standard floats will be the default going forwards. Toys can stay in the pram

[QMESAR]

Ben
Thanks that is good to hear as myself and the Team has been frustrating each other with the ICD for a few years and now we are at a stage where ICD is working including Ghost functions ,I do not want to walk this road a second time

.