[Spice-devel] About decisions and reviews

[Marc-André Lureau]

Hi

On Thu, May 14, 2020 at 2:32 PM Marc-André Lureau <
marcandre.lureau at gmail.com> wrote:

> Hi
>
> On Mon, May 11, 2020 at 12:16 PM Marc-André Lureau <
> marcandre.lureau at redhat.com> wrote:
>
>> Hi,
>>
>> About "Move code to C++":
>> https://gitlab.freedesktop.org/spice/spice/-/merge_requests/62
>>
>> I would like to know how the decision happened. As long as I have been
>> involved in the Spice project, we had open discussions and did
>> mandatory review for anything non-trivial.
>>
>> This change is substantial, and impacts the work and contribution of
>> others. Where did the discussion happen? Who reviewed the code
>> changes?
>>
>>
> Since no real discussion nor public review took place, I propose to revert
> the change and add a HACKING file to make the rules clear. See:
>
> https://gitlab.freedesktop.org/spice/spice/-/merge_requests/91
>
>
Based on the discussion there, and the fact that a majority of developper
are willing to take the change, I closed the MR.

Fwiw, beside the fact that the change wasn't done with review nor
discussion (which on its own is detrimental), I think using C++ is wrong:
- C++ is much more complex than C, and fewer developper know it. A good C++
developper is an even better C developper.
- in our common ecosystem, QEMU, Linux, the GNOME desktop, are 90% C
projects, and perhaps? 1% C++? (I don't know a single C++ system/core
project tbh - only apps like KDE/LibreOffice/Firefox)
- C++ projects are switching away from it (ex Firefox), I don't know of a
core/system project talking about switching to C++ these days.
- switching to C++ is going to be costly, as it is endless, and it will
leave a bitter taste of previous/existing C. We could have used that time
for other things... I would be pleasantly surprised if such refactoring
would lead to regressions.
- C has better interoperability story than C++, especially because the code
was GObjectified. We could have added/exported more friendly API/bindings
for GNOME easily.

Overall, it's a waste of effort to me. What is the overall direction of the
project? Except a few language improvements for a lower number of
developpers, what does C++ bring us that we couldn't do with C for our
users?

Nevertheless, I will try to invest some time reviewing the changes.

I hope we can agree on better project rules and directions for the future.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.freedesktop.org/archives/spice-devel/attachments/20200528/d89c2a68/attachment.htm>