Sjoerd de Vries

Module API bug

1

```python def do_superposition(...): ... ctx.do_superposition = Module() ctx.do_superposition.code = do_superposition ``` `ctx.do_superposition` now becomes a Transformer

probably the same in Jupyter. At some point, a prompt kit stdout patch gets installed. For now, try a fix by direct-printing to sys.__stdout__. Not sure how well this will...

It seems you have to first `del` the old instance.

"Synthetic" deep cells are deep cells whose elements are synthesized by the workflow, as opposed to being available in a remote database. Their members may be deleted from buffer cache...

Seamless graphs as self-consistent descriptors of computation results (without re-doing work)

1

Seamless graphs as containers of computation results ==================================================== UPDATE: the big data branch (elision + expression cache) has partially solved this. Seamless graphs contain the checksum of the computation results....

Triggered by passing a mapping context with a mount in it to map_dict_chunk

Module API needs work

2

1. To select a part of a Module and use it as another module. See the use of nefertiti in the RRM project. Requires that the submodule is independent, i.e....

This is to support FAIR queries later. Seamless will never query the database help cache, only write to it. Only synthesize and send it to database_sink upon save_graph. - Key:...

Non-caching transformer pins / direct_print_ cache misses in compiled transformers

1

Currently, Transformer.debug.direct_print is passed on to the executor of compiled transformers. This causes direct_print to become part of the transformation, leading to cache misses. Python, bash and bashdocker transformers are...