On 2016-12-01 09:19, Adrien Destugues wrote:
I'm not sure what great feat of engineering is needed to switch to gcc5h.
We literally already have it and a wide set of packages.
I'm looking for ways to attract new developers. I've had someone tell me
"we don't care about users who walk away from Haiku due to gcc2".
That is the worst possible position to have on this stuff.
That's not what I said, and I think you are referring to me.
what I said is, if people don't spend the 30 seconds it takes to
discover that we have:
- gcc5 based 64bit builds
- gcc5 based 32 bit builds, both with and without gcc2 apps support
Then it is unlikely that they will take time to write bug reports,
join the community, or be useful in any other way to the project. I
don't care about such "dead weight" users. I care about people who are
interested enough in Haiku to disccover such things and be useful
members of the community.
At this point in time, and given our current velocity, we have
all the time in the world for our next release. The numbers are out
there, and the numbers show a project that has entered maintenance mode:
https://www.openhub.net/p/haiku
"Activity" all the bursts of work and large undertakings are gone.
We aren't ever going to "finish" R1 in maintenance mode.
We are never going to finish R1 if people start pushing a lot of stuff
to trunk and there is no stable branch, either.
As soon as we go beta, the R1 branch will be in maintenance mode. It
seems I will be one of the very few people caring about it. Several
devs have already said they have no interest in R1 and want to work on
the exciting new R2 features. I see no problem with that, and I see
that "final" R1 may never happen. Still, I think there is room for a
"stable" branch in maintenance mode. We'll see if the availability of
an R2 branch makes development speed up again, that's up to you and
people not busy with the R1 branch maintenance.
I'm not trying to be a drama queen, but it is painful to watch us
slowly slide to irreverence after all the blood,sweat,and tears
we all have put into Haiku. The last few years have felt like a slow
glide of a luxury car into a block of concrete at a test track.
So we're going to get R1 out in 2016 or 2017?
Help welcome!
The main blocking point is still setting up the package repos so we
can create an R1 branch and have it use a different set of packages
than the R2 branch. I have solved most of the issues with the
haikuports recipes. Now, we need:
- More buildbots for the packages, especially if we want to support
gcc5h or x86_64 for beta1 (all bots currently online are gcc2h).
- Someone to set up the repos on Haiku servers (I'm not admin there
and don't want to be).
With that solved, we can create the R1 branch, point it to these new
repos, and then let the trunk open to all the R2 stuff. I don't know
if that will make it more active, we'll see. On the R1 branch we can
work on fixing or hiding the most annoying bugs and go for a release.
It won't be perfect and bug-free, but hey, it's only a beta.