-
Notifications
You must be signed in to change notification settings - Fork 130
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Same codebase on multiple client operating systems #310
Comments
Weep. Windows support is such a pita. Can you dump a diff somewhere, I want to know what paths are generated wrongly. |
Yeah, I can grab a diff in a few hours. I didn't look too closely so I'm not sure how many of them are paths and how many are line endings; I know the SConscript had the path issue and one of the other files had swapped line endings. For the paths in SConscript, I'd say the most obvious solution would be to have it do the path comversion at runtime: Python has functions to handle that if needed, so you can generate a single script that works on both and uses only a single path string representation in the source. |
Here's a diff from running lbuild on Linux after originally generating the files on Windows. Pretty much all of it could be fixed by standardizing on linux-style paths, I think, since forward slashes work on both systems while backslashes only work in that way on Windows. |
Completely agreed! There was a reason why the SConscript had to be this way, probably because Python doesn't properly handle Windows paths. - abspath(r"ext\cmsis\core"),
- abspath(r"ext\cmsis\device"),
+ abspath(r"ext/cmsis/core"),
+ abspath(r"ext/cmsis/device"), That's probably because the submodules differ. ST changes the permissions every now and again and I didn't catch on to it soon enough.
Newlines? Encoding? Wat? --- a/mcb-2019-2020-project/modm/ext/printf/printf.source
+++ b/mcb-2019-2020-project/modm/ext/printf/printf.source That's just weird? Maybe an ordering issue? diff --git a/mcb-2019-2020-project/modm/src/modm/architecture/interface.hpp b/mcb-2019-2020-project/modm/src/modm/architecture/interface.hpp
index cc3c7c3..cfa8163 100644
--- a/mcb-2019-2020-project/modm/src/modm/architecture/interface.hpp
+++ b/mcb-2019-2020-project/modm/src/modm/architecture/interface.hpp
@@ -14,6 +14,17 @@
#include <stdint.h>
+#include "interface/accessor.hpp"
+#include "interface/accessor_flash.hpp"
+#include "interface/accessor_ram.hpp"
+#include "interface/assert.hpp"
+#include "interface/atomic_lock.hpp"
+#include "interface/clock.hpp"
+#include "interface/delay.hpp"
+#include "interface/gpio.hpp"
+#include "interface/interrupt.hpp"
+#include "interface/memory.hpp"
+#include "interface/register.hpp"
#include "interface/peripheral.hpp"
#endif // MODM_INTERFACE_HPP
\ No newline at end of file Ok, this is just plain stupid, lbuild exposes the build log of the children modules to generate this list of header files and returns the native Python path for the platform. diff --git a/mcb-2019-2020-project/modm/src/modm/math.hpp b/mcb-2019-2020-project/modm/src/modm/math.hpp
index 802dc9b..88eb490 100644
--- a/mcb-2019-2020-project/modm/src/modm/math.hpp
+++ b/mcb-2019-2020-project/modm/src/modm/math.hpp
@@ -12,14 +12,14 @@
#ifndef MODM_MATH_HPP
#define MODM_MATH_HPP
-#include "math\tolerance.hpp"
-#include "math\units.hpp"
-#include "math\utils.hpp"
-#include "math\utils\arithmetic_traits.hpp"
-#include "math\utils\bit_constants.hpp"
-#include "math\utils\bit_operation.hpp"
-#include "math\utils\crc32.hpp"
-#include "math\utils\endianness.hpp"
-#include "math\utils\misc.hpp"
-#include "math\utils\operator.hpp"
+#include "math/tolerance.hpp"
+#include "math/units.hpp"
+#include "math/utils.hpp"
+#include "math/utils/arithmetic_traits.hpp"
+#include "math/utils/bit_constants.hpp"
+#include "math/utils/bit_operation.hpp"
+#include "math/utils/crc32.hpp"
+#include "math/utils/endianness.hpp"
+#include "math/utils/misc.hpp"
+#include "math/utils/operator.hpp"
#endif // MODM_MATH_HPP
\ No newline at end of file Realistically I need a Windows CI to fix this otherwise I'm just going to break Windows support again and again. |
Oh, this is just gold 🤦♂: headers = env.generated_local_files(filterfunc=
lambda path: re.match(r"interface/.*\.hpp", path)) |
I very much like that idea. Both to run a build in the first place, and ideally to try comparing a Linux build and a Windows build to make sure they have the same output. (although, that last one would require a service with pipelines and artifacts; not sure if the project has that right now.) |
I've added a partial fix to these problems in 457e815. This at least fixes the backslash issue in the aggregate header files. |
The templates for the SConscript file and a variety of other files are intentionally designed to produce paths (and, in some sporadic cases, newlines) which are specific to the OS the files are generated on.
I've been urged in a few other issues to check in the files that modm generates, but as a result if I generated my files on a Windows machine and then cloned the repo on a Linux machine, the project is unusable.
In my case, the following files are different between the two:
What am I supposed to do about this?
The text was updated successfully, but these errors were encountered: