[oi-dev] ruby 3.0.2

Till Wegmueller toasterson at gmail.com
Mon Sep 13 23:57:10 UTC 2021


While I can see your logic, I am afraid there wont be a package unless 
we make one. For one that I doubt Hashicorp makes one without us asking, 
and on the other hand it is not that easy to build IPS packages outside 
OI-userland. Which defines a lot of build time variables and structure. 
Also all of the people needed to make a review and get the software 
tested are in this community.

In any case a build from source Article will help. Can you make that?

If you are still up to make a OI package I would definitely help testing it.

-Till

On 13.09.21 15:37, stes at PANDORA.BE wrote:
> 
> Ruby 3.0.2 builds on OpenIndiana Hipster:
> 
> $ /scratch/ruby-30/bin/ruby --version
> ruby 3.0.2p107 (2021-07-07 revision 0db68f0233) [i386-solaris2.11]
> 
> The following config line works to build ruby 3.0.2:
> 
> ./configure --disable-dtrace --with-gcc --prefix=/scratch/ruby-30 BISON=/scratch/bison-35/bin/bison
> 
> This is with a custom bison instead of the currently OI installed bison 3.7.6.
> 
> There's some gar (gnu ar) errors when dtrace is enabled.
> 
> The gem ext/ripper grammar has a problem with bison 3.7.6 , the generated ripper.c is generated by the Ruby developers using bison 3.5.1.
> 
> However when I try to use vagrant with ruby 3.0.2 that poses a problem because the bundle --binstubs exec step from:
> 
> https://github.com/hashicorp/vagrant
> 
> is not working.  While with the tradition ruby-2.7.4 it works fine.
> 
> There may be more experimental features in ruby-3.0.2 compared to ruby-2.7.4 so 2.7.4 is preferable anyway.
> 
> So for the moment probably it just needs some testing (or a lot of testing) before IPS packaging this.
> 
> In fact "vagrant" can be used perfectly well from source ...
> 
> There could just be a community documentation page on docs.openindiana.org that documents how to install vagrant from source.  I'll make a PR for oi-docs to document this.
> 
> If a 'vagrant' IPS package is to be made, this would be more logically something for hashicorp,
> since it is hashicorp's product.
> 
> Regards,
> David Stes
> 
> _______________________________________________
> oi-dev mailing list
> oi-dev at openindiana.org
> https://openindiana.org/mailman/listinfo/oi-dev
> 



More information about the oi-dev mailing list