Discussion page for GSoC-2025 (Mukul)

[Mukulyadav2004]

@tdhock @Anirban166 @joshhwuu
Hi mentors, I’ll be using this thread to share regular updates throughout the coding period.

My end-semester exams are now wrapping up, and I’ll be starting working on project from now. I’ve spent some time during the bonding period exploring relevant parts of the data.table codebase, especially areas related to the initial issues outlined in my proposal.

I’ll begin by reviewing those issues in more detail. I’ll share regular updates here starting tomorrow.

[tdhock]

thanks!
In addition to writing on issues/PRs every day, I would also encourage you to write a blog every week which summarizes your progress and achievements.
see my gsoc links here for examples https://tdhock.github.io/teaching/

[Mukulyadav2004]

@tdhock @Anirban166 @joshhwuu

June 8, 2025 (Week 1 complete)
Submitted PR - Clarify x/i error prefixes in merge() for incompatible factor joins for #6641 to improve error messaging in merge() when x and i have incompatible factor levels. Though initially labeled as documentation, I proposed a fix by modifying bmerge.R and merge.R to raise a clearer and more accurate error directly instead of documenting the current behavior.

Also drafted a combined documentation update for #2002 and #2487, which clarify that first() and x[1] are distinct in purpose and behavior. I’ve shared the draft explanation on the issue page for feedback before proceeding with a PR - if anyone has suggestions or feedback on the approach, that would be helpful.

Also, I’ve updated my blog with the Week 1 progress and added the link here as well: Blog
I’ll keep posting updates there every weekend.

[tdhock]

looks great thanks!

[Anirban166]

Great start, keep it up!

PS: I saw a minor formatting issue while reading the post - Insert a newline after triple backticks / before first(integer(0)) on line 60 here

[Mukulyadav2004]

@tdhock @Anirban166 @joshhwuu

Week 2
This week, I focused on #4259 and started work on #6720.

• #4259 aimed to improve performance of as.IDate() for POSIXct inputs, especially with non-UTC time zones. I explored adding an as.IDate.POSIXlt method and using it internally. While the approach worked functionally, benchmark results showed a noticeable slowdown for large vectors, so we’ve decided not to pursue it further.
• For #6720, I’ve started drafting a central documentation file for all data.table options. I’m also working on adding an automated CI step to code-quality.yml, which will flag any option changes that aren’t reflected in the docs. The idea is to keep this list up to date automatically and reduce future maintenance friction.

If there’s anything I could be approaching differently or more effectively, I’d really appreciate any feedback or suggestions.

[Mukulyadav2004]

Hi mentors, just a quick update for today — I’ve prepared a .Rd file listing all data.table options, as part of work on #6720. You can view the result of the documentation getting on RStudio using the link below:

data.table options documentation

Let me know if you have any suggestions on formatting or content. Will continue refining and working on the CI step next.

[tdhock]

thanks! for the options, please submit a PR.

[Mukulyadav2004]

Hi mentors
I saw need of up to date list of all data.table options from issue discussion of #6720 so I prepared a separate file which include all data.table options along with their default and actual meaning and their respective purpose. After that to ensure the docs file to be up to date - I prepared a ci script that checks if all options present in .R file also included in our docs and vice versa.
But when I update workflow to include step to ran our script automatically then it is failing, can you please guide on where I'm doing wrong.

And could you please review them and confirm if I'm on the right track or its not worth doing.
Related PR #7075

[tdhock]

this makes sense, not sure why it is failing, but let's keep the discussion in that pr.

[Mukulyadav2004]

Hi @tdhock
I've just updated the blog with the progress from this week.

Apologies for not posting updates over the last two weeks - I spent quite a bit of time exploring issues that either turned out to be low-impact or not feasible to implement, so I didn't have solid progress to share during that time. Thanks for your patience!

[Mukulyadav2004]

Hi Mentors,
I hope you're doing well!
I've started diving into one of the next big tasks, which is data.table issue #5913 about creating classed conditions. It seems like the most important and challenging part is designing a good, consistent naming convention for the new classes.
Since this is a core design decision, I would really appreciate your guidance and perspective on it. I'd love to hear any initial thoughts you have on what makes a good class name in this context.
Thanks, and looking forward to discussing it with you!

[Mukulyadav2004]

Hi mentors!
I've updated the blog, kindly have a look.
Blog

[tdhock]

great thanks

[Mukulyadav2004]

Hi mentors! can you please review my understanding of this issue - #4567
Would appreciate your thoughts on it.

[Mukulyadav2004]

Hi @tdhock @Anirban166 @joshhwuu

Now that the midterm evaluation has concluded, I wanted to kindly ask for your feedback on my performance from your perspective.