This may be what you had in 
mind: 
https://github.com/gwtproject/old_google_code_wiki/blob/master/DistributedBuilds.wiki.md

You could also use a custom PermutationWorkerFactory to automate that, that 
would send the necessary files to another machine, run the CompilePerms 
there, then get the result 
back: 
https://github.com/gwtproject/old_google_code_wiki/blob/master/JavaSystemPropertiesAndGwt.wiki.md#compiler

Another solution could be to serialize the permutations' compilation rather 
than parallelize them by using a low localWorkers value 
(http://www.gwtproject.org/doc/latest/DevGuideCompilingAndDebugging.html#DevGuideCompilerOptions).
 
Build will be longer but put less pressure on memory.
You could also use fewer forked JVMs and more in-process permutation 
workers by bumping the gwt.jjs.maxThreads system property (if you set it to 
the same value as  -localWorkers or higher, everything will be done in the 
same JVM, without ever forking a new one)

Disclaimer: I've never had to deal with such builds, so can't be of much 
more help.


On Monday, February 15, 2021 at 8:41:39 PM UTC+1 Jon Kenoyer wrote:

> We are looking to expanding our GWT client to 13 plus locales in addition 
> to our current 5. The current code base takes 16 gigs to compile per 
> permutation (!). I remember at one point I found a blog post/discussion 
> topic where someone had put in an experimental feature to scale workers 
> across machines.
>
> Has anyone successfully done this and if some give some pointers?
>

-- 
You received this message because you are subscribed to the Google Groups "GWT 
Users" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to [email protected].
To view this discussion on the web visit 
https://groups.google.com/d/msgid/google-web-toolkit/7f7af224-e1a8-4329-8ac6-7d07dcee5883n%40googlegroups.com.

Reply via email to