Site: add Iceberg Rust and DataFusion Comet blog post#17162
Site: add Iceberg Rust and DataFusion Comet blog post#17162mbutrovich wants to merge 13 commits into
Conversation
| description: Apache Iceberg PMC | ||
| avatar: assets/images/iceberg-logo-icon.png | ||
| mbutrovich: | ||
| name: Matt Butrovich |
There was a problem hiding this comment.
This is the first non-Apache Iceberg PMC authored article, so I'd be establishing precedent here. Please let me know if you have any concerns here.
There was a problem hiding this comment.
Probably fine? This is just for auto-filling in the byline right?
There was a problem hiding this comment.
Yep. Mostly just making sure we didn't want any sort of standardized author format for non-PMC posts (i.e., putting some other sort of blanket author title with an author section in the body or something).
Sounds like what we've got will work.
There was a problem hiding this comment.
makes sense to include author for technical blog posts. past blogs were mostly announcements and used the iceberg pmc alias
| mbutrovich: | ||
| name: Matt Butrovich | ||
| description: Apache Iceberg Contributor, Apache DataFusion PMC Member, Apple | ||
| avatar: https://github.com/mbutrovich.png |
There was a problem hiding this comment.
We might want to make these stable links inside of the Iceberg repo, but didn't want to presume to upload my own photo just yet.
There was a problem hiding this comment.
TIL! and yea if we can avoid extra files, lets do that 😄
Kurtiscwright
left a comment
Kurtiscwright
left a comment
There was a problem hiding this comment.
Wonderful article and helped me understand what was mentioned at the Rust community sync.
I think this fits because it's specific to Iceberg development and explains enough about Comet to cover "How" this is done without the "What" or "Why's" having their focus shifted away from Iceberg.
I'm curious what the community thinks about where we draw the line for strictly non-iceberg projects being mentioned in these blog posts.
| We are thrilled by the deepening collaboration between the Iceberg and DataFusion communities, and we | ||
| encourage anyone interested to find a way to get involved. | ||
|
|
||
| ## Getting Involved |
There was a problem hiding this comment.
I think this is great to have as part of any Iceberg blog!
There was a problem hiding this comment.
I basically adapted it from the DataFusion blog posts in the first Iceberg Rust release blog post we had, and yeah I agree it should be a (possibly customized to the specific post's content) consistent footer on the blog.
| is a small but growing example of what that looks like in practice: two communities building on | ||
| each other's strengths to accelerate Iceberg on both fronts. Users who query Iceberg get faster |
There was a problem hiding this comment.
I like the positive tone this strikes.
The nitpick I have is the article hit me as how the Comet community is improving the Iceberg community, but the ending here reads more in the tone of both communities adding/informing changes to each other.
Does this article need a section or some refinement on how Iceberg is improving the correctness of Comet? Once again just a pedantic nitpick.
I think that's part of this process that we want to refine. Comet had accepted Iceberg Summit talks in 2026 and 2025, so it seems within scope on that precedent. But that doesn't necessarily mean every Iceberg Summit topic makes sense for a blog post, or that talks that didn't fit in Iceberg Summit shouldn't be a post here. I'm just raising it as a calibration point. I'm also happy to adjust the tone in specific directions based on feedback. Right now I think it strikes a nice balance of "building with Iceberg" and "building within Iceberg" but I'm always happy for more fresh eyes. Thanks for your feedback so far, @Kurtiscwright! |
|
Thanks for the feedback so far @Kurtiscwright and @RussellSpitzer! Let me address the comments this morning and put a new preview PDF in the description. |
|
taking a look, super excited to see more technical blogs. Thanks for writing it up! |
kevinjqliu
left a comment
kevinjqliu
left a comment
There was a problem hiding this comment.
thanks! this looks great, I have a few nit comments to tease out what i think readers should take away from this post. LMK what you think!
hsiang-c
left a comment
hsiang-c
left a comment
There was a problem hiding this comment.
Thanks Matt, this is a great post!
laskoviymishka
left a comment
laskoviymishka
left a comment
There was a problem hiding this comment.
Really enjoyed this article!
Approachable without dumbing anything down, and it stays anchored to Iceberg even while explaining a non-Iceberg project. The framing of "planning held constant, execution varies" makes the differential-testing insight land cleanly, and the fallback-relaxation-as-TDD-loop is a genuinely nice way to explain it.
Numbers check out (the ~40% and 102/103 both reconcile), and the figures are all captioned with alt text. Left one nit inline about a Spark docs version.
LGTM. 🚢👍
7 participants
@kevinjqliu encouraged me to write a blog post highlighting how DataFusion Comet 1) accelerates Spark queries over Iceberg tables and 2) accelerates Iceberg Rust development. I think this is an interesting blog post for both Iceberg users and developers. Since this is the first post of its type on the Iceberg blog, I'd like feedback on tone, technical scope, etc., in the context of our target audience. I have tried to keep it approachable. I'd also like feedback on figure readability, etc. I'm super excited about more writing on the Iceberg blog that reflects how the community engages with the project, and I'm hoping to be a guinea pig for more work like this from others.
I have a rendered preview here:
Accelerating Apache Spark Queries (and Iceberg Rust Development) with Apache DataFusion Comet.pdf