spark-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Matei Zaharia <matei.zaha...@gmail.com>
Subject Re: Proposal for Spark Release Strategy
Date Thu, 06 Feb 2014 20:07:24 GMT

On Feb 6, 2014, at 11:56 AM, Evan Chan <ev@ooyala.com> wrote:

> The other reason for waiting are things like stability.
> 
> It would be great to have as a goal for 1.0.0 that under most heavy
> use scenarios, workers and executors don't just die, which is not true
> today.
> Also, there should be minimal "silent failures" which are difficult to debug.
> 

I think this is orthogonal to the version number. 1.x versions can have bugs — it’s almost
unavoidable in the distributed system space. The version number is more about the level of
compatibility and support people can expect, which I think is something we want to solidify.
Calling it 1.x will also make it more likely that we have long-term maintenance releases,
because with the current project, people expect that they have to keep jumping to the latest
version. Just as an example, when we did a survey a while back, out of ~100 respondents, all
were either on the very latest release or on master (!). I’ve had multiple people ask me
about longer-term supported versions (e.g. if I download 1.x now, will it still have maintenance
releases a year from now, or will it be left in the dust).

Matei


Mime
View raw message