Fix dm_paste() pipe length issue by implementing operation chunking

[Copilot]

This PR addresses the issue where dm_paste() generates extremely long pipe chains that can cause C stack overflow errors and are difficult to read and debug.

Problem

When working with large data models containing many tables and relationships, dm_paste() would generate a single pipe chain with hundreds of operations:

dm(...) %>%
  dm_add_pk(table1, id) %>%
  dm_add_pk(table2, id) %>%
  ... # potentially 500+ operations
  dm_add_fk(table500, main_id, main)

This could trigger Error: C stack usage 7970324 is too close to the limit with older R/magrittr versions and creates unwieldy code that's hard to debug.

Solution

The fix implements automatic pipe chunking when the number of operations exceeds a configurable threshold (default: 100). Large pipe chains are split into manageable chunks with intermediate variable assignments:

dm_step_1 <- dm(...) %>%
  dm_add_pk(table1, id) %>%
  ... # first 100 operations

dm_step_2 <- dm_step_1 %>%
  dm_add_fk(table51, main_id, main) %>%
  ... # next 100 operations

dm_step_2 %>%
  dm_add_fk(table151, main_id, main) %>%
  ... # remaining operations

Key Changes

• Modified dm_paste_impl() to collect and count all operations before formatting
• Added dm_paste_chunk_operations() function to handle the chunking logic
• Maintains full backward compatibility - small dm objects work exactly as before
• Works with both magrittr (%>%) and base R (|>) pipe operators
• Added comprehensive test coverage for edge cases

Testing

The implementation has been thoroughly tested with:

• Edge cases (0, 100, 101, 250 operations)
• Both pipe operators (%>% and |>)
• Large dm objects with 500+ tables
• Integration with existing test suite

Fixes #400.

---

💬 Share your feedback on Copilot coding agent for the chance to win a $200 gift card! Click here to start the survey.