This definitely looks like something we'd want to investigate. Please
create an issue if you can:

https://groovy.apache.org/#reporting-issues

Or I'll do that shortly.

Cheers, Paul.

On Tue, Nov 4, 2025 at 5:32 AM Piotr Żygieło <[email protected]> wrote:
>
> Hello Groovy
>
> I wonder if what I'm facing (4c below) is working as expected:
>
> 1. download Groovy 5.0.2
> 2. unpack it to some GROOVY_HOME
> 3. place java/c (17, 21, 25) on PATH
> 4.
> a. use empty directory:
>   mkdir /tmp/empty && cd /tmp/empty
>   ${GROVY_HOME}/bin/groovysh
> This works fine.
>
> b. go to some other directory which is not empty, but with only
> several hundred files/directories:
>   ${GROVY_HOME}/bin/groovysh
> This works relatively fine.
>
> c. go to some other directory, like $HOME, or even $HOME/.m2 (really
> populated) - with a lot of files:
>   ${GROVY_HOME}/bin/groovysh
> This is only killable on my machine. (3.0.23 and 4.0.26 at least react
> to ^C, 5.0.2 - seems to be unterminatable).
>
> I guess it's something fancy in o.a.g.groovysh.Main or 
> o.c.g.tools.GroovyStarter
>
> https://github.com/pzrep/groovy-startup-time/actions/runs/19046142519/job/54394718799?pr=6
> shows it on hardware different than mine:
> - with timestamps enabled, time between the last sh's exec and grovysh
> prompt increases with number of files
> - while with relatively clean CWD - it takes 7 seconds for groovysh to
> start, just as few as 8 clones of openjdk/jdk repository make it more
> than 1 minute
>
> I find this very surprising.
>
> --
> Piotrek

Reply via email to