Repository navigation
Introduction: Nanotec is using KickCAT in a fork, and would like to contribute #456
Replies: 3 comments 1 reply
|
Hi Michael, |
|
Hi Michael, Glad that this project is of any help for you :) About the FoE::opcode renaming: this is indeed a public API change but since FoE is still considered experimental in KickCAT, I see no reason to not do this one. BTW, I consider it 'experimental' because so far I didn't have the chance to put my hand on a compliant/certified slave that can do FoE, so the whole implementation is documentation based and test only through our own slave stack (which can only prove that the code can do something, not that this is compliant). If possible, a PR with a MSVC CI job would be also a great addition so we don't break the project again on this compiler. Best regards, Philippe |
|
Hi Philippe, hi @trns1997, thank you both for the quick and friendly answers. I will open the pull requests in small steps:
An MSVC CI job makes sense. We currently build the KickCAT sources with our About FoE: we will use it against our own drives (e.g. firmware update). Best regards, |
Uh oh!
There was an error while loading. Please reload this page.
Hi all,
my name is Michael Milbradt, I am a software developer at Nanotec Electronic GmbH
near Munich, Germany. Nanotec develops and manufactures motors and motor controllers
for industrial automation. One of our software products is NanoLib, a C++ library
(with C#, Python and Java bindings) that our customers use to commission and control
our drives over CANopen, Modbus, EtherCAT and other fieldbuses.
We are planning to replace the current EtherCAT master stack in NanoLib with KickCAT. For now
this concerns the master side only; we have no plans for the slave stack. First of all:
thank you for this project. The code base is clean and easy to follow, and the
emulation layer makes it possible to test a master without hardware.
To do this we maintain a public fork:
https://github.com/nanotec-electronic/se-KickCAT
Why a fork: on Windows we build KickCAT with MSVC, not with MinGW, and that needs a few
changes to the sources. On Linux we use GCC, as upstream does. We also need a pinned,
reproducible revision for our releases. The fork is organised so that it stays close to
upstream:
masteris an unmodified mirror, every change lives on its own topic branch,and
nanotec/mainonly merges those. All our modifications to KickCAT files arepublished there under CeCILL-C.
We do not want to drift away from upstream. Fixes and improvements of general use we
would like to offer as pull requests. Two small ones exist so far:
KICKCAT_PACKED_BEGIN/ENDmacro pairinstead of
__attribute__((__packed__)), plusstatic_asserts on the structure sizes,and a few name collisions with Windows API macros.
PRId64instead of%ldfor nanosecond counts (longis32 bit on Windows).
One question before I open the first PR: the MSVC change renames
FoE::opcode::ERRORto
FoE::opcode::ERR, becauseERRORis a macro in<windows.h>(same reasoning asthe existing
mailbox::ERR). This is a public API change. Would that be acceptable, orwould you prefer a different solution?
If you have preferences on how contributions should look (PR size, tests, coding
style), please let me know and we will follow them.
Best regards,
Michael
All reactions