Remove Sampler and its interface

[penelopeysm]

what

As title. Essentially, this deletes the entirety of src/sampler.jl.

why

Because the type Sampler is gone, and also because DynamicPPL has no more actual samplers (in the past there were SampleFromPrior() and SampleFromUniform() that were actually samplers) that means we don't need to declare generic methods for AbstractMCMC.sample in DynamicPPL. That also means that the supporting interface used to define e.g. keyword argument defaults can also go.

Of course, it doesn't go completely: it just gets shifted upwards to Turing. In my opinion that makes a ton more sense anyway because that's where the concrete implementations of sample() are written. The current status is that these functions are defined in DynamicPPL but no methods are defined anywhere in DPPL, only in Turing. So may as well move the actual function there.

On top of this, the default implementation for sample was actually too restrictive in that it doesn't allow for AbstractMCMC.step_warmup to call DynamicPPL.initialstep. This means that you can't have a custom step_warmup for Turing samplers.

where now

My plan for which function goes where is outlined in the changelog:

DynamicPPL.jl/HISTORY.md

Lines 45 to 55 in be0d2ad

	### Removal of `DynamicPPL.Sampler`
	`DynamicPPL.Sampler` and **all associated interface functions** have also been removed entirely.
	If you were using these, the corresponding replacements are:
	- `DynamicPPL.Sampler(S)`: just don't wrap `S`; but make sure `S` subtypes `AbstractMCMC.AbstractSampler`
	- `DynamicPPL.initialstep`: directly implement `AbstractMCMC.step` and `AbstractMCMC.step_warmup` as per the AbstractMCMC interface
	- `DynamicPPL.loadstate`: `Turing.loadstate` (will be introduced in the next version)
	- `DynamicPPL.default_chain_type`: `Turing.default_chain_type` (will be introduced in the next version)
	- `DynamicPPL.initialsampler`: `Turing.init_strategy` (will be introduced in the next version; note that this function must return an `AbstractInitStrategy`, see above for explanation)
	- `DynamicPPL.default_varinfo`: `Turing.default_varinfo` (will be introduced in the next version)

For the upcoming release of Turing v0.41 I will just do a straight copy paste of the code here, instead of doing any larger refactoring. But I would like this PR to be in DPPL 0.38 so that we don't have to cut a new minor version just to delete this.

[github-actions]

Benchmark Report for Commit 35d0be2

Computer Information

Julia Version 1.11.6
Commit 9615af0f269 (2025-07-09 12:58 UTC)
Build Info:
  Official https://julialang.org/ release
Platform Info:
  OS: Linux (x86_64-linux-gnu)
  CPU: 4 × AMD EPYC 7763 64-Core Processor
  WORD_SIZE: 64
  LLVM: libLLVM-16.0.6 (ORCJIT, znver3)
Threads: 1 default, 0 interactive, 1 GC (on 4 virtual cores)

Benchmark Results

|                 Model | Dimension |  AD Backend |      VarInfo Type | Linked | Eval Time / Ref Time | AD Time / Eval Time |
|-----------------------|-----------|-------------|-------------------|--------|----------------------|---------------------|
| Simple assume observe |         1 | forwarddiff |             typed |  false |                  9.8 |                 1.4 |
|           Smorgasbord |       201 | forwarddiff |             typed |  false |                655.7 |                49.0 |
|           Smorgasbord |       201 | forwarddiff | simple_namedtuple |   true |                417.1 |                56.3 |
|           Smorgasbord |       201 | forwarddiff |           untyped |   true |               1068.5 |                33.1 |
|           Smorgasbord |       201 | forwarddiff |       simple_dict |   true |               8124.1 |                23.7 |
|           Smorgasbord |       201 | reversediff |             typed |   true |               1424.9 |                30.8 |
|           Smorgasbord |       201 |    mooncake |             typed |   true |                991.9 |                 4.6 |
|    Loop univariate 1k |      1000 |    mooncake |             typed |   true |               5847.2 |                 4.2 |
|       Multivariate 1k |      1000 |    mooncake |             typed |   true |                999.7 |                 8.9 |
|   Loop univariate 10k |     10000 |    mooncake |             typed |   true |              65432.2 |                 3.9 |
|      Multivariate 10k |     10000 |    mooncake |             typed |   true |               8741.3 |                10.0 |
|               Dynamic |        10 |    mooncake |             typed |   true |                145.4 |                11.7 |
|              Submodel |         1 |    mooncake |             typed |   true |                 13.9 |                 5.1 |
|                   LDA |        12 | reversediff |             typed |   true |               1044.2 |                 2.5 |

[github-actions]

DynamicPPL.jl documentation for PR #1037 is available at:
https://TuringLang.github.io/DynamicPPL.jl/previews/PR1037/

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 81.13%. Comparing base (e4c5bbd) to head (a3969f7).
⚠️ Report is 1 commits behind head on breaking.

Additional details and impacted files

@@             Coverage Diff              @@
##           breaking    #1037      +/-   ##
============================================
+ Coverage     80.98%   81.13%   +0.15%     
============================================
  Files            42       40       -2     
  Lines          3765     3722      -43     
============================================
- Hits           3049     3020      -29     
+ Misses          716      702      -14     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[sunxd3]

LGTM

[penelopeysm]

Thanks @sunxd3! Will hold off on merging until we're happy with TuringLang/Turing.jl#2676 so as to not pull the rug from under it...

[sunxd3]

Absolutely, I was going to mention this, but thought you had better judgement on the ordering of things. So no problem at all.