# OCaml Users Survey: 2023 vs 2026 Question Differences

**2023: 57 questions | 2026: 59 questions**

## Summary of Changes

- **New sections:** AI/LLM Usage (4 questions), Debugging & Profiling (3 questions)
- **New standalone questions:** Community size preference (Q17), CPU architectures (Q25), Documentation tools (Q37), Tooling feedback (Q38), Package repository feedback (Q47)
- **Removed questions:** 10 questions removed
- **Net change:** +2 questions

---

## New Sections

### Community Growth (Q17)
New question about community size preferences.

Q17. Regarding the size of the OCaml community, I wish the community was...
- Smaller
- The current size is about right
- Bigger
- A lot bigger
- I have no opinion

*Justification: Understanding whether the community wants growth helps inform outreach and adoption strategies.*

### AI/LLM Usage (Q18-Q21)

Four new questions about AI/LLM tool usage for OCaml development.

Q18. Do you use AI/LLM tools for OCaml development?
Q19. Which AI/LLM tools do you use with OCaml?
Q20. What do you use AI/LLM tools for in OCaml development?
Q21. Anything else you want to say on AI/LLM tools or use for OCaml development?

*Justification: AI tools have become prevalent in software development since 2023; understanding adoption patterns helps prioritize tooling investments.*

*Implementation note: This section can be made conditional in Google Forms. If Q18 = "No, ..." options, questions Q19-Q21 can be skipped.*

### CPU Architectures (Q25)

New question about target CPU architectures.

Q25. Which CPU architectures do you target?
- x86_64
- ARM64 (aarch64)
- ARM32
- RISC-V
- POWER
- S390
- I don't know
- Other

*Justification: Provides data for compiler team resource allocation, especially for RISC-V demand and ARM32 relevance (Android Wearables/TV).*

### Debugging & Profiling (Q39-Q41)

Three new questions focused specifically on debugging approaches and profiling tools.

Q39. What debugging approach do you primarily use?
Q40. Which debugging and profiling tools do you use?
Q41. What are your biggest challenges with debugging/profiling?

Q40 includes: ocamldebug, earlybird, gdb/lldb, ppx_minidebug, Logs library, perf, eBPF (Linux), DTrace (FreeBSD/macOS), memtrace, Landmarks, magic-trace.

*Justification: 2023 free-text responses frequently cited debugging difficulties as a pain point, particularly lack of good debugging for native executables.*

### Documentation Tools (Q37)

New question about documentation tooling.

Q37. Which documentation tools do you use?
- odoc
- ocamldoc
- None
- Other

*Justification: Understanding documentation tool adoption helps prioritize ecosystem investments.*

---

## Removed Questions

| 2023 Q# | Question | Reason for Removal |
|---------|----------|-------------------|
| Q18 | How large was the largest OCaml project you contributed to? | Low actionability; project size data didn't yield actionable insights |
| Q27 | How do you feel about the compiler's every-six-month release schedule? | Low actionability; responses consistently skewed to "just right" |
| Q31 | Where do you host your software sources? | Not OCaml-specific; GitHub dominance (87%) made this uninformative |
| Q35 | If you deploy on the web, what platform(s) are you using? | Low response rate; most respondents don't deploy web applications |
| Q38 | OCaml tooling provides a comfortable workflow for me | Consolidated—Q45 (tooling confusing to newcomers) captures the key insight |
| Q40 | What would you change? [about tooling] | Replaced with specific text field Q38 ("What would you change about OCaml tooling?") |
| Q42 | What would you change? [about package repos] | Replaced with specific text field Q47 ("What would you change about OCaml's package repositories?") |
| Q46 | OCaml libraries are stable enough for my needs | 69% satisfaction in 2023; stability is not a priority concern |
| Q48 | Software written in OCaml is easy to maintain | 74% agreement in 2023; maintainability is a known strength, not a concern |
| Q51 | If I was granted up to three new language features... | Effect handlers shipped in OCaml 5; question is outdated |

---

## Modified Questions

### Q3. Usage context
- **Added options:** "For teaching", "For work, in open source"
- *Justification: 2023 responses showed ~10 teaching-related "Other" entries; open-source work (e.g. funded maintainers) is distinct from industry or hobby*

### Q4. OCaml usage at workplace
- **Added option:** "I don't use OCaml at work"
- *Justification: Respondents who are employed but only use OCaml for personal projects had no appropriate option*

### Q8. Other programming languages
- **Added:** Zig
- *Justification: Growing language in the systems/FP space*

### Q9. Software types developed
- **Added:** "Command-line tools/utilities", "Games", "Mobile applications (Android/iOS)"
- *Justification: 2023 "Other" responses showed CLI tools and Games as common categories; mobile development is present in Q10 domains but was missing here*

### Q11. How did you first hear about OCaml?
- **Changed:** From select dropdown to text field
- *Justification: 2023 showed diverse discovery paths; free text captures nuance better*

### Q15. Community interaction platforms
- **Added:** Bluesky, Telegram, "Local meetups / user groups"
- **Changed:** "X (formerly Twitter)" → "X / Twitter"
- *Justification: Platform evolution since 2023; IRL meetups were missing from options*

### Q23. Development platforms
- **Added:** WebAssembly
- **Added clarification:** "(Your development machine, not cross-compilation targets.)"
- *Justification: 2023 "Other" responses included WebAssembly; clarification avoids confusion with cross-compilation targets*

### Q24. Target platforms
- **Added:** "Web/Browser (via js_of_ocaml or Melange)", WebAssembly, "Unikernels (MirageOS)"
- *Justification: 2023 "Other" responses included these; WebAssembly support has improved; Melange is now a major web compilation target alongside js_of_ocaml*

### Q26. Language implementations
- **Added:** OxCaml
- *Justification: OxCaml (Jane Street's OCaml fork) is publicly available and gaining visibility*

### Q27-Q28. Compiler versions
- **Changed:** Collapsed old versions (4.02-4.06 into "≤4.07")
- **Added:** 5.2, 5.3, 5.4, 5.5
- *Justification: Only 2% used versions older than 4.08 in 2023; OCaml 5.x adoption is now mainstream*

### Q29. Installation methods
- **Added:** "dune package management"
- **Changed:** "Npm" → "npm/yarn/pnpm"
- *Justification: Dune package management is new since 2023; npm alternatives are common*

### Q31. Upgrade pain points
- **Split:** "Breaking changes" into "Breaking changes in the type system" and "Breaking changes in the standard library"
- **Added:** "Reliance on a compatible ppxlib version"
- *Justification: More specific breakage data helps compiler team; ppxlib compatibility is a known bottleneck distinct from general dependency issues*

### Q32. Build tools
- **Added:** "Nix-based builds"
- *Justification: Nix usage for builds is distinct from Nix for package management*

### Q33. Preprocessors
- **Clarified:** "cpp-style conditional-compilation directives" → "cpp-style conditional-compilation directives (e.g. cppo)"
- *Justification: The original wording was vague; cppo is the most common tool in this category*

### Q34. Editors
- **Added:** Helix, Kakoune, Zed
- **Changed:** Split "Vim (or vim-based such as Neovim)" into separate "Neovim" and "Vim" options
- *Justification: Emerging editors with OCaml/LSP support; Neovim has a distinct plugin/LSP ecosystem from Vim*

### Q36. Benchmarking tools
- **Changed:** From generic categories to specific tool names
- **2023:** "Microbenchmarking library (core_bench, benchmark, bench...)", "Performance-monitoring libraries (landmark...)", etc.
- **2026:** core_bench, benchmark, bechamel, perf, Landmarks, memtrace, magic-trace, "Whole-program timing (time, hyperfine)", "Custom project-specific tools", None
- *Justification: Specific tools enable better ecosystem understanding*

### Q42. Deployment methods
- **Added:** "As WebAssembly modules", "As a unikernel"
- *Justification: WebAssembly deployment is increasingly relevant; unikernel deployment (MirageOS) was listed as a target platform but missing from deployment options*

### Q54. State-of-the-art ecosystem wish
- **Added options:** "AI/LLM tooling for OCaml", "I don't know"
- *Justification: Reflects current ecosystem priorities; "I don't know" was missing for respondents without a strong preference*

### Q55. Adoption pain points
- **Added option:** "Poor AI/LLM tool support compared to other languages"
- *Justification: AI tooling quality increasingly affects language adoption decisions*

### Q56. Final comments
- **Changed:** Removed 255 character limit
- *Justification: Character limit was unnecessarily restrictive*

---

## Unchanged Questions (Preserved for Comparability)

The following questions are identical or functionally equivalent to enable year-over-year comparison:

| Topic | 2023 | 2026 | Notes |
|-------|------|------|-------|
| OCaml experience duration | Q1 | Q1 | Identical |
| Proficiency self-assessment | Q2 | Q2 | Identical |
| Workplace OCaml trend | Q5 | Q5 | Identical |
| How learned OCaml | Q6 | Q6 | Identical |
| School/University name | Q7 | Q7 | Identical |
| Domains of software | Q10 | Q10 | Identical |
| Learning pain points | Q12 | Q12 | Identical |
| OCSF awareness | Q13 | Q13 | Identical |
| OCSF priorities | Q14 | Q14 | Identical |
| Content topics wanted | Q16 | Q16 | Identical |
| Project contribution count | Q17 | Q22 | Identical |
| Testing approaches | Q32 | Q35 | Identical |
| Feel welcome in community | Q36 | Q43 | Identical |
| Satisfied with language | Q37 | Q44 | Identical |
| Tooling confusing to newcomers | Q39 | Q45 | Identical |
| Satisfied with package repos | Q41 | Q46 | Identical |
| Can find libraries | Q43 | Q48 | Identical |
| Missing libraries | Q44 | Q49 | Identical |
| Libraries well documented | Q45 | Q50 | Identical |
| Understand best practices | Q47 | Q51 | Identical |
| Can find OCaml jobs | Q49 | Q52 | Identical |
| Can find OCaml candidates | Q50 | Q53 | Identical |
| Country | Q55 | Q57 | Identical |
| Underrepresented group | Q56 | Q58 | Identical |
| Difficulty participating | Q57 | Q59 | Identical |

---

## Question Number Mapping

| Section | 2023 | 2026 |
|---------|------|------|
| OCaml Usage | Q1-Q12 | Q1-Q12 |
| OCSF | Q13-Q14 | Q13-Q14 |
| Community | Q15-Q16 | Q15-Q17 |
| AI/LLM Usage | — | Q18-Q21 |
| Projects & Contributions | Q17-Q20 | Q22-Q25 |
| Compiler | Q21-Q27 | Q26-Q31 |
| Tooling | Q28-Q33 | Q32-Q38 |
| Debugging & Profiling | — | Q39-Q41 |
| Infrastructure | Q34-Q35 | Q42 |
| Feelings | Q36-Q50 | Q43-Q53 |
| Burning Desires | Q51-Q54 | Q54-Q56 |
| Demographics | Q55-Q57 | Q57-Q59 |

---

## Detailed Question Mapping (2023 → 2026)

| 2023 | 2026 | Status |
|------|------|--------|
| Q1 | Q1 | Unchanged |
| Q2 | Q2 | Unchanged |
| Q3 | Q3 | Modified (added "For teaching", "For work, in open source") |
| Q4 | Q4 | Modified (added "I don't use OCaml at work") |
| Q5 | Q5 | Unchanged |
| Q6 | Q6 | Unchanged |
| Q7 | Q7 | Unchanged |
| Q8 | Q8 | Modified (added Zig) |
| Q9 | Q9 | Modified (added CLI tools, Games, Mobile) |
| Q10 | Q10 | Unchanged |
| Q11 | Q11 | Modified (select → text field) |
| Q12 | Q12 | Unchanged |
| Q13 | Q13 | Unchanged |
| Q14 | Q14 | Unchanged |
| Q15 | Q15 | Modified (added platforms) |
| Q16 | Q16 | Unchanged |
| — | Q17 | **New** (community size) |
| — | Q18-Q21 | **New** (AI/LLM section) |
| Q17 | Q22 | Unchanged |
| Q18 | — | **Removed** |
| Q19 | Q23 | Modified (added WebAssembly, clarification) |
| Q20 | Q24 | Modified (added targets, Melange) |
| — | Q25 | **New** (CPU architectures) |
| Q21 | Q26 | Modified (added OxCaml) |
| Q22 | Q27 | Modified (collapsed old versions) |
| Q23 | Q28 | Modified (collapsed old versions) |
| Q24 | Q29 | Modified (added dune pkg mgmt) |
| Q25 | Q30 | Unchanged |
| Q26 | Q31 | Modified (split breaking changes, added ppxlib) |
| Q27 | — | **Removed** |
| Q28 | Q32 | Modified (added Nix-based builds) |
| Q29 | Q33 | Modified (clarified cppo) |
| Q30 | Q34 | Modified (added editors, split Vim/Neovim) |
| Q31 | — | **Removed** |
| Q32 | Q35 | Unchanged |
| Q33 | Q36 | Modified (specific tools) |
| — | Q37 | **New** (documentation tools) |
| — | Q38 | **New** (tooling feedback text) |
| — | Q39-Q41 | **New** (debugging section) |
| Q34 | Q42 | Modified (added WebAssembly, unikernel) |
| Q35 | — | **Removed** |
| Q36 | Q43 | Unchanged |
| Q37 | Q44 | Unchanged |
| Q38 | — | **Removed** |
| Q39 | Q45 | Unchanged |
| Q40 | — | **Removed** (replaced by Q38) |
| Q41 | Q46 | Unchanged |
| Q42 | — | **Removed** (replaced by Q47) |
| — | Q47 | **New** (package repo feedback text) |
| Q43 | Q48 | Unchanged |
| Q44 | Q49 | Unchanged |
| Q45 | Q50 | Unchanged |
| Q46 | — | **Removed** |
| Q47 | Q51 | Unchanged |
| Q48 | — | **Removed** |
| Q49 | Q52 | Unchanged |
| Q50 | Q53 | Unchanged |
| Q51 | — | **Removed** |
| Q52 | Q54 | Modified (added AI/LLM option, I don't know) |
| Q53 | Q55 | Modified (added AI/LLM option) |
| Q54 | Q56 | Modified (removed char limit) |
| Q55 | Q57 | Unchanged |
| Q56 | Q58 | Unchanged |
| Q57 | Q59 | Unchanged |