DavidSpickett wrote: > That was just a symptom, but what we want is users not having to specify > process launch -w at all. We want our users to simply do run and that things > just work.
We have: * process launch -w does not carry the working dir to subsequent launches. * Please open an issue for that. * Maybe it can be fixed with the setting, but you don't need to deal with that. * Your users want to use `run` as they would normally, I think I would too. * Adding a setting is a reasonable approach for this part. * It's kinda like `git -C` but for all targets launched by lldb. You will have to decide whether `process launch -w` or the setting wins, I would say `process launch -w` wins and if `-w` is not passed then the setting is used. > launch-working-dir sounds like working-dir to me. Although one might think that `working-dir` actually changes the working directory of a running target if you change it at runtime. We already have a few settings that actually only apply on first launch but aren't named as such. So let's leave the `launch` bit in and document that changing it at runtime will not change the process' working dir, to be extra clear. > Should this be target.process not target? On second thought, if this is on target then the launch workdir dir should be too: ``` target.run-args -- A list containing all the arguments to be passed to the executable when it is run. Note that this does NOT include the argv[0] which is in target.arg0. ``` ...so ignore advice from me on a Wednesday I guess :) Also you could reference how this setting interacts with the other ways to pass arguments like `lldb program -- arg1 arg2`, whether it's additive or one wins. For a working dir, one has to win of course. https://github.com/llvm/llvm-project/pull/113521 _______________________________________________ lldb-commits mailing list lldb-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/lldb-commits
