bobwire / pentoo

Automatically exported from code.google.com/p/pentoo
1 stars 0 forks source link

dev-ruby/rspec-its missing ~> rake-10.1.0 dependency #315

Open GoogleCodeExporter opened 9 years ago

GoogleCodeExporter commented 9 years ago
Note: this is bottom priority, from bleeding-edge operation.

Neither the Pentoo or Gentoo tree has rake-10.1.x, but the rspec-its package 
(all versions) has an internal dependency on it (see 
https://github.com/rspec/rspec-its/blob/master/rspec-its.gemspec).

Where this affects Pentoo is that pentoo/pentoo-exploit pulls in www-apps/beef, 
which through a dependency on dev-ruby/em-websocket eventually depends on 
dev-ruby/rspec-its.  This breaks updates in an easily-solved "--keep-going" 
manner.

Specifically it appears that dev-ruby/em-websocket-0.5.x introduces this 
dependency tree, so sticking with the 0.3.x branch by masking makes for a 
"cleaner" fix.

Original issue reported on code.google.com by aoz....@gmail.com on 15 Nov 2014 at 8:55

GoogleCodeExporter commented 9 years ago
the following commit was completely wrong (as few others):
https://code.google.com/p/pentoo/source/detail?r=6094

We need to review each ebuild and check depences.

ps. ZC revoke his access.

As for now, please make sure to use our profile. beef does not pull 
dev-ruby/em-websocket-0.5.x here.

I've also specified it in the ebuild, see r6176

Original comment by blshkv on 16 Nov 2014 at 3:13

GoogleCodeExporter commented 9 years ago
That's fine - I see that and will unmask it locally.

I am using the Pentoo profile, as the system was grown from one of the 2014 RC 
cycle:
[root@box ~] eselect profile show
Current /etc/portage/make.profile symlink:
  pentoo:pentoo/default/linux/amd64

I just have ACCEPT_KEYWORDS set, which obviously steps a little farther out of 
bounds than the bleeding_edge, profile.  I'm still fine with the occasional 
breakage this causes.

Original comment by aoz....@gmail.com on 16 Nov 2014 at 4:02

GoogleCodeExporter commented 9 years ago
Actually, r6176 introduces a bit of breakage itself; only 
dev-ruby/uglifier-2.5.3 is in either tree, not 2.2 as the second part of the 
change requires.

Original comment by aoz....@gmail.com on 16 Nov 2014 at 4:09

GoogleCodeExporter commented 9 years ago
yeah, thanks. The gemfile wants "uglifier", "~> 2.2.1" which basically mains 
2.2*. But let's just use 2.5 because 2.2 is just OLD and i see no point in 
forking it...

Original comment by blshkv on 16 Nov 2014 at 12:36

GoogleCodeExporter commented 9 years ago
btw, ACCEPT_KEYWORDS is good to verify if the right version specified in the 
ebuild..
And this is exactly how I fixed the ebuild. All available version will be 
unmasked for you with this keyword, however ebuild will pull the only allowed 
version i.e. 0.3.

Feel free to open more bugreport is if you find something similar. Cheers.

Original comment by blshkv on 16 Nov 2014 at 12:41

GoogleCodeExporter commented 9 years ago

Original comment by blshkv on 16 Nov 2014 at 12:42

GoogleCodeExporter commented 9 years ago
ok, the following two commits need to be reviewed:
https://code.google.com/p/pentoo/source/detail?r=6094
https://code.google.com/p/pentoo/source/detail?r=6088

Some ebuilds like em-websocket "use Gemfile for development purposes" (see 
https://github.com/igrigorik/em-websocket/issues/131). Instead, gemspec file 
should be used.

So the copy-pasted trick with "BUNDLE_GEMFILE=Gemfile ${RUBY} -S bundle install 
--local" can't be used.

and yeah, ruby sucks.

Original comment by blshkv on 24 Nov 2014 at 11:18