> After almost ten years of development, we have the regret to announce that we will put an end to our Theano development after the 1.0 release, which is due in the next few weeks. We will continue minimal maintenance to keep it working for one year, but we will stop actively implementing new features.
I disagree.
Theano is in widespread use and released under a permissive license.
The software will stay around. The contributions from LISA/MILA have clearly been crucial historically, but we don't know at this point what will happen to Theano when they direct their attentions elsewhere.
Personally I am inclined to think people and development will move to the other libraries Bengio alludes to, but even so, spreading a public declaration of the "death" of a pioneering decade-long project is not to be done lightly. Spreading that kind of rumors is a contribution towards making sure Theano doesn't have a future after MILA.
> Summarizing this as "dead" is not hyperbolic IMO.
Theano is a Python library. Software doesn't die once someone decides that the design and implementation stages are over and the project enters the maintenance stage. The software remains operational and in use, and if bugs are found they will be squashed. That's it.
This is nearly the same announcement that came from Deis at the last community planning meeting at the start of August. In fact, Microsoft only promised 6 months of maintenance after their final v2.18 Deis Workflow release.
And you know, before the first Thursday of September, when the final release was made and it would have been time for the next community planning meeting, three companies had already stepped up and committed to continue development.
People and companies are still using Deis, for better or for worse, and although the Deis core team are elsewhere working on more Kube-native toolings, the core "Workflow" tooling is now fixed in place and for some of us, its functionality is essentially complete. For others, development of the tools will surely continue as no vaguely similar competitor exists, and those teams' needs may have not even been imagined yet.
Do you know what I've found? It is really significantly easier for a casual like myself to work on Deis now that it is in maintenance; you can actually commit yourself to making a change that will take a good long while, and with your own limited resources, when you know that the platform has ceased to be a moving target. You don't need to track everyone's concurrent development efforts to keep your own contributions relevant, if those other development efforts have stopped.
If this project is important to you, then you should pledge to support it in whatever material way you can. The project team has committed to keep the lights on for you for at least a year, so if you ever wanted to be a maintainer for an open source project, there is no time like the present. So rally the troops and step up to the plate!
6 comments
[ 3.8 ms ] story [ 15.5 ms ] threadI'm not sure why this post seems fail merging automatically but the topic has been posted recently and received some attention.
https://news.ycombinator.com/item?id=15358586
https://news.ycombinator.com/item?id=15360470
> After almost ten years of development, we have the regret to announce that we will put an end to our Theano development after the 1.0 release, which is due in the next few weeks. We will continue minimal maintenance to keep it working for one year, but we will stop actively implementing new features.
Summarizing this as "dead" is not hyperbolic IMO.
Theano is a Python library. Software doesn't die once someone decides that the design and implementation stages are over and the project enters the maintenance stage. The software remains operational and in use, and if bugs are found they will be squashed. That's it.
And you know, before the first Thursday of September, when the final release was made and it would have been time for the next community planning meeting, three companies had already stepped up and committed to continue development.
People and companies are still using Deis, for better or for worse, and although the Deis core team are elsewhere working on more Kube-native toolings, the core "Workflow" tooling is now fixed in place and for some of us, its functionality is essentially complete. For others, development of the tools will surely continue as no vaguely similar competitor exists, and those teams' needs may have not even been imagined yet.
Do you know what I've found? It is really significantly easier for a casual like myself to work on Deis now that it is in maintenance; you can actually commit yourself to making a change that will take a good long while, and with your own limited resources, when you know that the platform has ceased to be a moving target. You don't need to track everyone's concurrent development efforts to keep your own contributions relevant, if those other development efforts have stopped.
If this project is important to you, then you should pledge to support it in whatever material way you can. The project team has committed to keep the lights on for you for at least a year, so if you ever wanted to be a maintainer for an open source project, there is no time like the present. So rally the troops and step up to the plate!