On Sat, 27 Apr 2019 at 15:58, Siddhesh Poyarekar <siddhesh@xxxxxxxxxx> wrote:
On 27/04/19 6:50 PM, Dibyendu Majumdar wrote:
Hi, I think that maintaining LuaJIT is a full time job ... are you
proposing to devote a lot of time to it going forward? It would be sad
if this was a temporary/short-term effort.
I agree it would be sad for me personally if I weren't able to carry
this forward full time and forever since it's a fun project to tinker
with, but it's not about me. I see myself only as an enabler in helping
revive the project by hopefully getting everyone running their own forks
to buy into the idea (and enable them) to contribute to a single repository.
So my idea is not to create another centre of command (I am not even
worthy of that given how little time I have spent on luajit compared to
others here) but to have multiple empowered developers who can run the
project together.
Maybe if I set a goal to add at least two more committers to my fork in
the next 6 months, it would be a better guarantee of continuity?
Secondly how do you propose to test all those patches, on all
supported platforms?
This is where getting patches into distributions and getting
LuaJIT-test-cleanup into the main repository (both of which I have done
with my repo) is key. Through Fedora I have test coverage of i686[1],
x86_64[2], s390x[3], armhf[4] and arm64[5]. Through Debian my test
coverage will hopefully be wider. In addition to that, I have set up CI
jobs[6][7] on the Linaro CI instance that triggers on every commit.
I'll be happy to help add CI nodes to the Linaro CI jobs; the Linux
Foundation Open Mainframe project has kindly offered an s390x instance
to add to it.
Of course, test coverage is one thing and fixing failures in tests
(armhf and i686 are currently failing for example and those failures are
without any of my patches to Luajit proper) is quite another thing. I
am not proposing to do this myself, which is why I posted here to get
help from others who have an interest in keeping the project alive.
