[Rpm-ecosystem] Announcing DNF 3 development
ngompa13 at gmail.com
Mon Mar 26 12:30:44 UTC 2018
On Mon, Mar 26, 2018 at 8:26 AM, Martin Sehnoutka <msehnout at redhat.com> wrote:
> On 03/26/2018 01:38 PM, Neal Gompa wrote:
>> On Mon, Mar 26, 2018 at 7:22 AM, Matěj Cepl <mcepl at cepl.eu> wrote:
>>> On 2018-03-26, 10:52 GMT, Florian Weimer wrote:
>>>> On 03/22/2018 01:40 PM, Daniel Mach wrote:
>>>>> Please read more details on our blog:
>>>> “C++ 11 is supported by GCC in RHEL 7 / CentOS 7” — You should
>>>> use Developer Toolset to compile on Red Hat Enterprise Linux
>>>> 7 if you need C++11 support. The system compiler, GCC 4.8,
>>>> has limited support only.
>>> When switching the programming langauge than I would think there
>>> are some better C-successors than C++, namely Rust? Mad rush of
>>> giving up on 46 years old language and switching to one which is
>>> just 33 years old seems a bit bizarre to me.
> Take a look into the code, it is mostly C with few features from C++.
> btw what is the motivation to use GOBjects? Is the libdnf api supposed
> to be consumed by dnf frontend via gi repository?
It was a thought a while ago with libhif, and as part of the final
rationalization for libdnf, it's being dropped. Because libdnf is
going to be in C++, it's going to use SWIG for bindings generation.
>> I'm okay with not dealing with LLVM for my system package manager,
>> thank you very much. I'd be more open to Rust if Rust also could be
>> built with GCC, and thus supported across literally everything, but no
>> one is investing in that effort.
> Well, investment like this will need some justification, not saying that
> dnf should be the one, but you will definitely need a big, important
Considering all the other "big important things" people don't invest
in anyway, I don't think that'd help any.
>> And frankly, Rust is harder to program in than C++, and creating
>> bindings is no walk in the park.
> Purely personal opinion. You are probably referring to the learning
> curve, which is known to be steep, but after this period it is well
> worth the effort.
Not my personal opinion. That's the opinion of several developers I
know who are working on Rust based projects. Not everyone gets the
benefit of GNOME forcing all the things so that stuff _must_ work.
> Regarding the bindings, if libdnf is meant to be used via gir (see my
> question above), then there is already an effort to make this much
> easier (I'm referring to gnome-class).
As I noted earlier in this email, gir is a leftover and is being removed.
真実はいつも一つ！/ Always, there's only one truth!
More information about the Rpm-ecosystem