On Sat, 30 Sep 2017 16:36:05 +0000 ng0 wrote: > Christopher Baines transcribed 3.6K bytes: > > On Sat, 30 Sep 2017 15:01:52 +0000 > > ng0 wrote: > > > > > ng0 transcribed 2.1K bytes: > > > > Christopher Baines transcribed 1.7K bytes: > > > > > On Sat, 30 Sep 2017 14:12:55 +0000 > > > > > ng0 wrote: > > > > > > > > > > > The fix in this version is to only add the necessary input > > > > > > to the inherited gnutls. > > > > > > > > > > Ok. One hopefully final thing. From the commit message [1], > > > > > it's not clear to me if this is fixing an issue with the > > > > > GNUnet package, by providing it with the right dependencies, > > > > > or, adding additional functionality to the GNUnet package, by > > > > > providing a more capable GnuTLS? > > > > > > > > > > 1: "GNUnet and its dependency chain needs GnuTLS with DANE > > > > > support." > > > > It provides the right GnuTLS to GNUnet, libmicrohttpd and gnURL. > > > > Certain features of these applications will not work without it. > > > > GnuTLS without Dane is not fatal error for these packages, but > > > > Dane is recommended. > > > > I have no idea how I should put this into the very strict > > > > dictionary we have in commit messages… In more free-form it > > > > would be no problem for me. > > > > -- > > > > ng0 > > > > GnuPG: A88C8ADD129828D7EAC02E52E22F9BBFEE348588 > > > > GnuPG: https://krosos.org/dist/keys/ > > > > https://www.infotropique.org https://krosos.org > > > > > > Okay, I now see what you mean. > > > > > > To make it short: the correct dependency is provided in place > > > of the current working-but-not-correct GnuTLS. > > > > > > Is that more clear? > > > > I think you've made it clear to me now. > > > > From what you're saying, I think its the 2nd thing I said. Making > > this change will enable some functionality in the GNUnet package > > (and possibly some of the other packages changed). > > > > I'm not sure using the word "correct" helps, unless you say what the > > dependencies are correct/incorrect with respect to, for example, if > > the GNUnet documentation says that it should be built with GnuTLS > > with Dane support, then that would be a reason to talk about > > correctness. > > Yes it does state this, in the documentation and all these > applications list the dependency in README aswell. > This is why I thought the original, first version, comment I made in > gnutls/dane was enough. But I'll have to be more clear then. > Well if gnURL doesn't state it in its README I have to add it, if it's > not in there it's my mistake - it's correct nevertheless. Ah yes, I didn't really take in that comment in when I read the patch, but I can see it now "This is required for GNS functionality by GNUnet and gnURL.". That's great information, and having it in the commit would also be great.