My biggest concern with a file lock is: what happens if the other process terminates unexpectedly and does not remove the file lock? How can we ensure this won't happen?
On Thu, Jan 27, 2022 at 9:47 AM Brendan Ball <brendan.voicewo...@gmail.com> wrote: > Hi > > We've been getting weird errors when building releases: > ``` > Compiling 42 files (.ex) > ** (File.Error) could not remove files and directories recursively from > "/home/brendan/dev/app/_build/prod/lib/shared_app/priv": file already exists > (elixir 1.13.2) lib/file.ex:1292: File.rm_rf!/1 > (mix 1.13.2) lib/mix/utils.ex:444: Mix.Utils.symlink_or_copy/3 > (mix 1.13.2) lib/mix/project.ex:738: anonymous fn/5 in > Mix.Project.build_structure/2 > (elixir 1.13.2) lib/enum.ex:2396: Enum."-reduce/3-lists^foldl/2-0-"/3 > (mix 1.13.2) lib/mix/project.ex:737: Mix.Project.build_structure/2 > (mix 1.13.2) lib/mix/tasks/compile.all.ex:34: > Mix.Tasks.Compile.All.run/1 > (mix 1.13.2) lib/mix/task.ex:397: anonymous fn/3 in Mix.Task.run_task/3 > (mix 1.13.2) lib/mix/project.ex:396: Mix.Project.in_project/4 > ``` > > We build different releases concurrently after running a single `mix > compile`. > It turns out that the different release builds were not running in > isolated environments, but shared the filesystem. This is a very weird and > confusing error especially since most devs didn't know why it was breaking > and didn't know that releases were being run in a shared environment. > > This seems like something that might happen fairly often in CI > environments, so I really think it's in everyone's interest to improve this. > Most good build systems/compilers implement some kind of file lock for > operations that can't be done concurrently. I think implementing this would > be a minimal acceptable solution. Ideally the file lock is used to wait > until the operation can be performed, but even just an error saying you > can't run operations in parallel would be better than what we have now. > > Thoughts? > > Kind regards > Brendan Ball > > > -- > You received this message because you are subscribed to the Google Groups > "elixir-lang-core" group. > To unsubscribe from this group and stop receiving emails from it, send an > email to elixir-lang-core+unsubscr...@googlegroups.com. > To view this discussion on the web visit > https://groups.google.com/d/msgid/elixir-lang-core/8dfd467c-ec43-40de-bb2d-c9632d3de0can%40googlegroups.com > <https://groups.google.com/d/msgid/elixir-lang-core/8dfd467c-ec43-40de-bb2d-c9632d3de0can%40googlegroups.com?utm_medium=email&utm_source=footer> > . > -- You received this message because you are subscribed to the Google Groups "elixir-lang-core" group. To unsubscribe from this group and stop receiving emails from it, send an email to elixir-lang-core+unsubscr...@googlegroups.com. To view this discussion on the web visit https://groups.google.com/d/msgid/elixir-lang-core/CAGnRm4JbnaeaiWXZ1fsFiEXHZLVU-%3DFXWva%2BLO9-vQshFMqi_g%40mail.gmail.com.