Kjetil H
kjetil-hartveit.com
Kjetil H
@kjetil-hartveit.com
Web developer and sports lover. Founder and developer of @fantasy-snooker.com. My personal website is old 😅
Seems like Maps are serializable by React so no need to fix those.

Apparently “use cache” doesn’t persist across builds so that’s a big cause for the many writes. Wonder if they’ll enable Data Cache for “use cache” later.

Might reduce use of generateStaticParams and use “use cache” more sparingly
January 2, 2026 at 2:27 PM
Discarded the excluding of updatedAt fields, not worth. Gonna find a sane solution for the “use cache” stuff and make sure all cached stuff is serializable (I’m returning some Map instances a few places) and then check if things improves
January 2, 2026 at 12:14 AM
I just want to build and do things the right way, but I don't know if I'm doing it right or not.

I've done `next build` twice in a row and try to compare the builds with an AI generated (obv) script. A bunch of `createServerReference` and `registerServerReference` diffs, so not sure if they count
January 1, 2026 at 11:40 PM
Happy new year! Next.js/Vercel doesn't want me to have nice things though bsky.app/profile/kjet...
I'm trying to understand why I have so many ISR writes with Next.js and Vercel and have created a script which compares two builds, I'm now excluding updatedAt fields with Prisma, moved "use cache" outside of components which took in promises, tried to stabilize outputs from functions
January 1, 2026 at 11:34 PM
Next.js told me I used new Date() in functions returning Prisma data (so lies) so I threw in a bunch of "use cache" and now my comparison script tells me I have like 500 more diffs.

I'm wondering what kind of hell this is. Why am I spending time on this and why aren't there better tools
January 1, 2026 at 11:31 PM
An earlier boss got me en.wikipedia.org/wiki/What_If.... Nice nerdy gift :)
What If? (book) - Wikipedia
en.wikipedia.org
November 14, 2025 at 4:06 PM
Nearly 4x improvement locally though..
May 25, 2025 at 7:01 AM
As mentioned in the issue there exists a rule called no-implicit -coercions which can disallow shorthands like +foo but then you would just change it to Number(foo) and have the same problem as before
March 16, 2025 at 9:20 AM