Re: [gentoo-dev] [PATCH] eclass/ruby-fakegem.eclass: guard against broken rdoc
On Fri, 2022-07-15 at 13:04 +0100, Sam James wrote: > > I worry a bit about silently skipping given it can go further > and lead to e.g. unpredictably broken binpkgs. Yes, not really happy with that either, even though this only affects documentation files. > I'd probably prefer not doing +doc but I assume it's there > for a reason. Your question prompted me to investigate this again. +doc was added initially to ensure that the built-in documentation was generated which people expect to work as part of normal operations. Starting with bundler 2.2.0 this documentation is now bundled as part of the gem and does not need to be generated anymore, so for the 2.2 and 2.3 series the +doc can be dropped. I'll drop the patch since I expect that this is no longer needed (or at least, the drawbacks now outweight the benefits. Hans signature.asc Description: This is a digitally signed message part
Re: [gentoo-dev] [PATCH] eclass/ruby-fakegem.eclass: guard against broken rdoc
> On 15 Jul 2022, at 11:17, Hans de Graaff wrote: > > Skip building documentation rather than generating an error when rdoc is > missing but documentation is requested. rdoc should not be missing > normally, but this is a common problem when updating from one ruby > target straight to another one. There can be a gap when eselect has not > been triggered again yet for rdoc and another core ruby package may > require rdoc. This is commonly bundler since it has a +doc USE flag. > I worry a bit about silently skipping given it can go further and lead to e.g. unpredictably broken binpkgs. I'd probably prefer not doing +doc but I assume it's there for a reason. Anyway, the situation is what it is until we get a ruby-exec or similar, so go for it. Thanks for figuring out a workaround, as this one is often a bit painful. > Signed-off-by: Hans de Graaff > --- Best,, sam signature.asc Description: Message signed with OpenPGP
