I recently went through the exercise of contributing my first PR to spark. 

While I think that the JIRA definitely contains a rich history and project 
context, I will add that the workflow was a bit novel compared to the classical 
“open an Issue and discuss” contribution workflow. 

I do think there may be some risk of bifurcation of project history from one 
platform to another, but I would also add that it would lower the barrier of 
entry for first time contributors. 

—Antonio

Sent from my iPhone. 

> On May 20, 2026, at 2:43 PM, Tian Gao via dev <[email protected]> wrote:
> 
> 
> Hi,
> 
> From our discussion 3 months ago 
> https://lists.apache.org/thread/kv11qlr8j05cwqjoyddybclwcn0nv2n7 , we've 
> decided to open github issues as an experimental option and evaluate the 
> results after 3 months. Now it's time!
> 
> In the last 3 months, we opened github issues solely for bug reports and 
> improvement requests. We got 63 issues in total. Most of them are valid, we 
> probably had 1 or 2 spams. Even without any template (it was never merged), 
> almost every issue was created with a detailed description. Very few issues 
> came from committers.
> 
> I (Claude Code) did a quick search in JIRA. In the same time frame, I asked 
> it to list JIRA tickets that:
>   1. is either bug or improvement
>   2. has a description longer than 50 words        
>   3. assignee is not reporter
>   4. either there is no linked PR, or the first PR was created more than 12 
> hours after the ticket was created
> 
> This filter attempts to exclude "community reports", ruling out JIRA tickets 
> created just for submitting a PR.
> 
> The number is approximately 47. I think it's safe to say that we received at 
> least the same amount of feedback from the community on GitHub as on JIRA. 
> Another interesting number is 1394 - that's the number of JIRA tickets 
> created during this timeframe. What it represents is probably beyond this 
> thread's discussion, but that's an interesting diff.
> 
> Now back to our next step. According to my proposal, we had to choose from 
> one of the following options:
> 
> 1. Explicitly declaring that we need more time for this experiment. 3 or 6 
> extra months.
> 2. Close the github issues because the maintenance effort is larger than the 
> benefit.
> 3. Decide that using github issues as discussion only is the best way for 
> spark and keep doing it.
> 4. Support github issues as an equivalent to JIRA tickets so PRs can link to 
> them too.
> 5. Fully migrate from JIRA to github issues.
> 
> Personally I think github issues have proven useful, but different opinions 
> are definitely appreciated. There's almost no maintenance cost for github 
> issues so I don't think we need to choose 2. I'd love to hear from the 
> community about their preferred option. To make it more objective, I won't 
> choose one here.
> 
> Tian

Reply via email to