Multiple calls to initRDKitModule

[fredludlow]

Reposting from #436

Firstly, thanks a lot for all the work that's gone into this project. It's really nice to have the rdkit functionality in the browser. I have a (hopefully) basic question, which might be answered in the docs somewhere but I couldn't find it.

If I have a webapp in which multiple independent parts might require rdkit, what's the convention around calls to initRDKitModule. Each call seems to trigger a fetch of the wasm code, so I'm guessing its setting up a new instance per-call.

I can easily define some convention within my app whereby the promise (return value of initRDKitModule) is saved on some global object or shared state, but I'm aware I might be missing something here.

[ptosco]

@fredludlow Hi Fred, as you already figured out, every time you call initRDKitModule() a new instance of the module will be created.
You may take a look at how the rdkit-structure-renderer package avoids multiple instantiations: Renderer.init() is safe to call multiple times, as it will only instantiate the RDKit module once.
You might as well find convenient to use rdkit-structure-renderer (it is a vanilla JS package with very few dependencies other than MinimalLib itself) rather than implementing the logic again in your app.
Note that rdkit-structure-renderer does not depend on rdkit-js - it bundles its own version of MinimalLib, which may include additional functionality compared to rdkit-js. rdkit-structure-renderer is built from the RDKit master branch, while rdkit-js follows the RDKit release cycle, hence the MinimalLib bundled in rdkit-structure-renderer is more bleeding edge than the one in rdkit-js.

[fredludlow]

Thanks @ptosco - I'm actually using it to enable client-side substructure querying rather than depiction - apart from the differences around release cycles/bleeding edge you mentioned, are there any other differences (e.g. does rdkit-structure-renderer exclude any functionality as not directly relevant to rendering structures?)

Somewhat relatedly, how do I configure the size of the emscripten/C++ side heap, and is there any way I can check how much is currently in use? (I'm aware I need to call delete() and I've not got hard numbers from profiling but I have a suspicion I could improve performance by caching the rdkit mol instances - it would be helpful to have a way to know memory usage).

[ptosco]

@fredludlow

are there any other differences (e.g. does rdkit-structure-renderer exclude any functionality as not directly relevant to rendering structures?)

No - rdkit-structure-renderer may rather include additional, optional functionality compared to rdkit-js; i.e., ``rdkit-structure-rendererwill always be a superset ofrdkit-js` rather than a subset.

Somewhat relatedly, how do I configure the size of the emscripten/C++ side heap, and is there any way I can check how much is currently in use?

To configure the size of the emscripten heap, you may use INITIAL_MEMORY, MAXIMUM_MEMORY and ALLOW_MEMORY_GROWTH, all of which can be configured with emcmake at build time. See for example this thread:
emscripten-core/emscripten#14667
To check how much memory is available from JS, you may call rdkitModule.HEAP8.length. Getting memory which is currently use can be more tricky; you might have a look at this thread for some pointers:
emscripten-core/emscripten#8082