-
Notifications
You must be signed in to change notification settings - Fork 343
Problems building Paradox.sln [beta04] #116
Comments
Similar problems for Paradox.Android.sln: BuildLog.txt.zip | uploaded via ZenHub ... and Paradox.iOS.sln: |
I noticed the problem with {0} displaying (which we will fix), but it should still display the exception info right after in the current version (if there is any) Other than that, I couldn't reproduce the exception/error you had with the version on GitHub. By any chance, can you try to set "SiliconStudio.Assets.CompilerApp" as your start project, and copy paste the command arguments from build log? (grep for "BuildEngine arguments"), then debug it with Exception Thrown (Ctrl+Alt+E) on? (please ignore Net/Socket related one). In your case, the arguments that you should debug with should be: |
Also interested to know what displays if you change https://github.com/SiliconStudio/paradox/blob/master/sources/assets/SiliconStudio.Assets.CompilerApp/Program.cs#L222 |
No exception is thrown (in the debug session). I added a --log-file argument to capture the output:
|
You sure "Thrown" column is checked when you do Ctrl+Alt+E? |
Positive. is returning a result with HasErrors true and then it exits cleanly (with BuildError exit code). |
I can trace it all the way to here: Call stack just before task run:
|
Great, thanks! |
Interesting, it seems to happen when compiling the required assembly (using MSBuild). |
The [redirected] log first has those lines after shelling out to MSBuild. i.e. after stepping past |
Ok I see what is going on: This works fine in user games, but not with unit tests because we don't have a shared game but it uses the exe itself for that (needed for pdxfx, asset plugins, etc...), which call the build engine recursively. Also, we usually skip that if done in Debug mode (we build that game assembly in debug mode, so we expect it to be built as expected in that case). I thought that was something that was fixed some time ago, but it seems it's not. Thanks for the help. |
FYI: I updated to commit ceba0b1. It took 3 attempts to build Paradox.sln. The first was from a clean checkout (of my fork, followed by a merge of ceba0b1). The second was incremental from there after having to cancel the first build which had hung indefinitely trying building the test projects. I had to cancel the second build which had also hung trying building the test projects. The third time it finally built after doing a clean before the build. Weird. |
Yes it can be closed, thanks! |
Some test projects don't build:
I'm using the latest source as at commit b47b609. The build is sufficiently complex that I can't make head or tail of it. ;)
BuildLog.txt.zip | uploaded via ZenHub
The text was updated successfully, but these errors were encountered: