Re:Cinder: Instagram's performance-oriented fork of CPython
by terminalcommand on Wednesday May 05, 2021 @08:43AM
(#27049156)
What more could we expect from them to do? They are a for-profit company. Opensourcing their entire interpreter stack with years of proprietary development is a bold move. It shows that they care about open source and want to give back to the community.
They could've kept these developments behind closed doors and nobody would mind it.
They also offered to provide support if anyone wanted to contact them. That is top-class behaviour from instagram, I personally couldn't expect more.
Re:Cinder: Instagram's performance-oriented fork of CPython
by black_puppydog on Wednesday May 05, 2021 @08:48AM
(#27049195)
Not saying what they did was not good; it's certainly better than keeping things closed, but to answer these two parts:
> What more could we expect from them to do?
> That is top-class behaviour from instagram
To be considered top-class, I would expect them to go the extra mile and contribute to upstream. As many other companies have done over the past and continue to do.
Re:Cinder: Instagram's performance-oriented fork of CPython
by _carljm on Wednesday May 05, 2021 @09:08AM
(#27049364)
We have contributed quite a few parts of cinder upstream already and will continue to do so. I’m not sure where you got the idea that this is just a dump and we expect core devs to do all the work of upstreaming.
Re:Cinder: Instagram's performance-oriented fork of CPython
by black_puppydog on Wednesday May 05, 2021 @10:36AM
(#27050386)
I'm very happy to hear that _carljm. I was just replying to my GP's "they took the 'dump it and see' approach" and my Parent's "What more could we expect from them to do?"
If you're going that mile, then my hat is off to you and the team, that's how it's supposed to be done! :)
In some ways a bit cheeky that they took the 'dump it and see' approach rather than offering to work to upstream it, since it's kinda outsourcing the work of maintaining the performance improvements to the Python core devs rather than offering to put some of Facebook's considerable resources towards doing it themselves.