fix: use the proper qualified struct tag as column cache key

[hperl]

No description provided.

[Copilot]

Pull request overview

This PR fixes a cache collision issue in the column cache by changing the cache key from using just the table name to using a fully qualified type name (package path + type name). This prevents different struct types that map to the same database table from incorrectly sharing cached column definitions.

Key Changes

• Added a new cacheKey() function that generates cache keys based on the fully qualified type name (PkgPath + Name)
• Updated buildColumns() to use the new type-based cache key instead of the table name
• Minor formatting changes to commented-out code for consistency

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[zepatrik]

Note: depending on the use-case, you can also use .Select(colNames...). But this caching is really a pita...

[hperl]

Note: depending on the use-case, you can also use .Select(colNames...). But this caching is really a pita...

If there are additional column names, it won't cache anyways

pop/sql_builder.go

Line 270 in f4c800e

if acl == 0 {

[zepatrik]

If there are additional column names, it won't cache anyways

Yes that's what I wanted to say 😅

[zepatrik]

Nice, now we can have multiple structs for one table 😍
The question remains, is this now actually improving anything, or could we just remove the cache in general?

[hperl]

Nice, now we can have multiple structs for one table 😍 The question remains, is this now actually improving anything, or could we just remove the cache in general?

Well the cache is useful in general, because reflection is somewhat slow, so caching the columns per struct is still nice.

[alnr]

Nice, now we can have multiple structs for one table 😍 The question remains, is this now actually improving anything, or could we just remove the cache in general?

Well the cache is useful in general, because reflection is somewhat slow, so caching the columns per struct is still nice.

Right, but now we use reflection to compute the cache key...