alejandro

@janodev
26 Followers
201 Following
67 Posts
iOS dev

Couple of curl scripts to let Claude ask o3/gemini when he becomes stuck.

https://gist.github.com/janodev/45493d4323b8cb6797a0688864288aa1

note to self

Xcode 16.3 builtin rsync misbehaving leads to ITMS-90048: Your archive contains [._Symbols]

https://developer.apple.com/forums/thread/776674?answerId=833073022#83307302

Solution: custom export and remove manually

# check
unzip -l YourApp.ipa| grep -i "._Symbols"

# delete
zip -d YourApp.ipa ._Symbols/

Then upload with Transporter https://apps.apple.com/es/app/transporter/id1450874784?l=en-GB&mt=12

This bundle is invalid - Your arch… | Apple Developer Forums

Claude Code can increase test coverage working unattended. I use this Makefile https://gist.github.com/janodev/03809990aa99e079d62353eaebf7b5ba for a SPM package. Claude is able to see the report and run isolated tests. It uses xcbeautify -q to minimize token usage. Unattended = I’m watching but often I don’t need to interact.
Clauding

Clauding. GitHub Gist: instantly share code, notes, and snippets.

Gist
Some LLM tips
– Use 'swift build | xcbeautify' in Claude Code to avoid walls of text.
– I find solving concurrency problems in a blog post and adding it to the context works great.
– LLMs can’t one shot, but you need to actively review their output.
– It’s useful to ask ”what would have helped you to know to solve this problem faster?”. Then add the info to the code, a blog post, or project doc to supplement LLM shortcomings and often for your own learning too.
PreferenceKey is always one overlapping state update from crashing the SwiftUI renderer. Rather not use it.
I’ve been involved in project estimation lately. It’s massively shorter using Claude Code, even at current price. It’s beyond me why Apple doesn’t use Deepseek to clone CC and give it away as an onsite free alternative to boost hardware sales and development on their platform.
Not vibe coding, just babysit and manual refactor. SwiftSyntax to slice source code in meaningful parts (done) and send it to openai ada (best for code embeddings). Then MCP for file manipulation. Just sharing the recipe, not selling anything, the moat is non-existent.