Raw h5py.Dataset attribute access from lazy objects

[vinamarora8]

Adds a context manager: temporaldata.resolve_on_access(enabled):
This allows accessing the raw h5py.Dataset attributes of lazy objects without loading them into memory. Examples:

import temporaldata as td

data = td.Data.load("test.h5")

with td.resolve_on_access(False):
    x = data.spikes.timestamps  # <- this will be an `h5py.Dataset` and not materialized

x = data.spikes.timestamps # <- this gets auto-loaded into a numpy array into memory

Alternative names: autoresolve(False), raw_access(True)

Some considerations:

• If the lazy object has pending lazy operations (slice, mask, etc.), then the return h5py.Dataset would not reflect those lazy ops, and a warning will be printed explaining this.

[coderabbitai]

Important

Review skipped

Draft detected.

Please check the settings in the CodeRabbit UI or the .coderabbit.yaml file in this repository. To trigger a single review, invoke the @coderabbitai review command.

⚙️ Run configuration

Configuration used: defaults

Review profile: CHILL

Plan: Pro

Run ID: 686723ae-8b67-405c-a714-0634934e377c

You can disable this status message by setting the reviews.review_status to false in the CodeRabbit configuration file.

Use the checkbox below for a quick retry:

• 🔍 Trigger review

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Commit unit tests in branch vinam/control-autoresolve

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands and usage tips.}

[codecov]

Codecov Report

❌ Patch coverage is 91.42857% with 3 lines in your changes missing coverage. Please review.

Files with missing lines	Patch %	Lines
temporaldata/arraydict.py	80.00%	1 Missing ⚠️
temporaldata/interval.py	80.00%	1 Missing ⚠️
temporaldata/regular_ts.py	83.33%	1 Missing ⚠️

📢 Thoughts on this report? Let us know!

[github-actions]

Benchmark comparison vs main

Baseline: origin/main
Target: HEAD

Running benchmarks for dedd91fa1d...
Running benchmarks for 29be9dea66...

  Benchmark                                     dedd91fa1d (µs)    29be9dea66 (µs)    Speedup
  --------------------------------------------------------------------------------------------
  Data.slice() (lazy, realistic)                       1880.361           1906.462      0.99x
  Data.slice() (in-memory)                              731.426            744.986      0.98x
  IrregularTimeSeries.slice()                            85.933             86.048      1.00x
  RegularTimeSeries.slice()                              43.526             43.789      0.99x
  Interval.slice()                                        9.316              9.339      1.00x
  Interval.__and__ (1k&single)                           50.179             50.096      1.00x
  Interval.__and__ (1k&100)                            1175.396           1176.341      1.00x
  Interval.__or__ (1k|100)                               65.322             64.240      1.02x
  Interval.difference (1k-100)                         5714.059           5736.926      1.00x
  ArrayDict.keys() x100k                                  1.812              1.780      1.02x
  LazyInterval access (10 attrs)                       2331.902           2327.398      1.00x

[vinamarora8]

CC: @ShivashriganeshMahato @divyansha1115 I remember you needed something like this for your projects. What do you think of this API?

[vinamarora8]

Also, @milosobral would love your thoughts too!

[divyansha1115]

CC: @ShivashriganeshMahato @divyansha1115 I remember you needed something like this for your projects. What do you think of this API?

Overall, this looks very useful, especially for large datasets. Regarding the API, passing a boolean feels a bit awkward and could easily lead to mistakes by end users. It might be more clear to provide two separate context managers instead (e.g., raw_access() and lazy_access()).