Le 27/08/2015 05:33, Michael Gilbert a écrit : > On Wed, Aug 26, 2015 at 7:11 PM, Linus Torvalds wrote: >> So quite frankly, the fact that Debian people now attack Dirk, who has >> been bending over backwards over these kinds of stupidities, is not a >> big surprise. I think it's in the Debian DNA to care more about rules >> than about technical sanity. The whole "this is the only right way to >> do licensing" thing extends to "this is the only correct waty to do >> packaging". >> >> It's sad to see. > Perhaps it isn't clear external to the project, but participation in > the Debian BTS does not automatically make one a Debian person, which > not a defined status. > > The maintainer of the subsurface package, an official Debian status, > removed the package on request, so in many ways, problem happily > solved? > Even if I am not the initial packager of subsurface, I was the maintainer for the past few years. I think we worked great with Dirk. He uplifted some of my patches without any issue, always been friendly and quick to answer. He was a great upstream. We quickly discussed at the New Orleans during Linux Plumber about this packaging.
Now, he thinks that the way we do things in Debian is not the right way for subsurface. To be honest, for libraries being heavily modified, I agree with him. Sometimes, it is necessary to ship embedded sources. It is a crazy waste of everybody time to try handling that. Daily, I see that at work. Being a release manager for a major web browser, I see a lot of hacks being done on some libraries like Cairo or Skia. Even if they get merged upstreams and upstreams release a new version, the latency is too important. That does not scale for Debian. Recently, a top crash on fedora was caused because they were using the system cairo and not the one shipped by upstream ( https://bugzilla.redhat.com/show_bug.cgi?id=1253086 ). In some cases, shipping third party libraries into the upstream code is the best way. Sylvestre
