Long delay after building #171
Comments
|
I'm using [lein-cljsbuild "0.2.10"]. |
|
I had the same problem with 0.2.10. I did as follows $ cd a-cljs-project and the hanging disappeared mimmo On Jan 11, 2013, at 11:52 AM, Wei Hsu wrote:
|
|
That didn't fix it for me. Wonder if it's related to a particular dependency. Here's my project.clj: |
|
and this is the project against which I tested lein-cljsbuild 0.2.10 https://github.com/magomimmo/modern-cljs.git HIH to check if it's a dependency related issue mimmo On Jan 11, 2013, at 1:02 PM, Wei Hsu wrote:
|
|
the same delay issue have been experienced again by me just by substituting domina 1.0.0 with 1.0.2-SNAPSHOT in my project dependencies. |
|
I don't think this is dependency-related -- I think it has to do with how Leiningen is pumping input/output to the process. Here is a stack trace of |
|
Issue #147 might have a hint as to the source of this problem: Based on that commit and the stack trace, I think this is somehow related to input pumping in Leiningen's |
|
This commit has a hack that seems to fix this problem (but I don't fully understand why): |
|
Thanks for addressing this Evan! Your solution might shed light on another hanging problem in one of my own projects. |
|
Sure thing. My solution is definitely a hack -- it works for me but I don't totally understand it. Hopefully it works for everyone... |
address emezeske#171. This caused a regression in testing and notify behaviour that was reported as emezeske#453, where subprocesses would hang for 30 seconds after the subprocess completed. This commit avoids the need for pump threads to copy data from subprocess streams by using ProcessBuilder's existing Redirect functionality to either write redirect stream output to files directly, or to inherit System.out and System.err from the main process. This means that `lein test` etc now terminate in a reasonable time again, and avoids the call to `exit` that was causing problems for trampoline tasks.
Pull request emezeske#436 removes the call to exit that was added to address emezeske#171. This caused a regression in testing and notify behaviour that was reported as emezeske#453, where subprocesses would hang for 30 seconds after the subprocess completed. This commit avoids the need for pump threads to copy data from subprocess streams by using ProcessBuilder's existing Redirect functionality to either write redirect stream output to files directly, or to inherit System.out and System.err from the main process. This means that `lein test` etc now terminate in a reasonable time again, and avoids the call to `exit` that was causing problems for trampoline tasks.
Pull request #436 removes the call to exit that was added to address #171. This caused a regression in testing and notify behaviour that was reported as #453, where subprocesses would hang for 30 seconds after the subprocess completed. This commit avoids the need for pump threads to copy data from subprocess streams by using ProcessBuilder's existing Redirect functionality to either write redirect stream output to files directly, or to inherit System.out and System.err from the main process. This means that `lein test` etc now terminate in a reasonable time again, and avoids the call to `exit` that was causing problems for trampoline tasks.
When I run
lein cljsbuild once, the process prints "Successfully compiled ... in 6.727668 seconds." and then waits a long time (about 30s on my machine) before returning. This means thatlein cljsbuild testtakes over a minute to run. What is the build process doing during that time?The text was updated successfully, but these errors were encountered: