CDeviceNameService - list devices about circle HOT 10 CLOSED

Hi Andrzej, I have added the suggested method. It can be called using

m_DeviceNameService.ListDevices (&m_Screen);

from CKernel::Run(). Block devices are preceded with a 'b' character. Thanks for this suggestion.

from circle.

I forgot to mention: it's on the develop branch.

from circle.

Thanks. Unfortunately i can't link circle develop (and circle-newlib) since you broke backward compatibility. I have to mamy libraries compiled with float-abi=hard.

from circle.

Can you please try to change the ARCH options in Rules.mk for your RASPPI model back to -mfloat-abi=hard ? Does it work then? This is still under development and would be valuable information. Perhaps another way would be to recompile all the libraries, if this is possible.

The problem is, that for the Linaro GCC 7.1.1 toolchain there isn't support for -mfloat-abi=hard for the RPi 1 any more. That's why I changed this setting, but perhaps I have to revert it or make it configurable.

from circle.

I have added a build option FLOAT_ABI to make it configurable. It defaults to the previous setting hard, but may be set to softfp if required. I think, this is the best solution.

from circle.

Hi. Thanks. I'm sure this setting is fine.

But i still can't link it with my libraries. Plenty of stupid errrors like

..\circle-stdlib\install\arm-none-circle\lib\libm.a(lib_a-w_acos.o)||In function acos':| w_acos.c:(.text+0xb4)||undefined reference to __errno'|
..\circle-stdlib\install\arm-none-circle\lib\libm.a(lib_a-w_sqrt.o)||In function sqrt':| w_sqrt.c:(.text+0xc8)||undefined reference to __errno'|

Damn, i'm hopeless.

from circle.

I have to apologise, because I invited you to do tests of versions from our development branches. Previously this should not be a problem, because versions, checked in into the Circle develop branch, have passed a module test and should work, normally. But we have a new situation now, that we are working together on different projects and sub-projects, using this branch for versions, which are not tested very well, to be able to communicate. This was not clear to me.

Because we had several commits yesterday, I do not know, which commits of the different projects you have used for your test and cannot reproduce your problem. I would say, it's better to sit back now and give us some days to sort this out and to come to a state, which makes sense for further testing. I will inform you here, when this has been reached.

from circle.

@etet100: Apparently you are building Circle in the context of the circle-stdlib project. There was an error in the Makefiles of the samples that the C library libc.a was put before the math library libm.a on the linker command line. This order was wrong because libm.a references symbols from libc.a, namely __errno.

This is fixed now, the error should no longer appear starting with commit smuehlst/circle-stdlib@a1ed3f7.

from circle.

Thank you very much! A found some more bugs in my code and now it's working fine.

from circle.

Great! Good to know, that it is already working fine. So I think, this can be closed.

from circle.