seamless
seamless copied to clipboard
Seamless is a framework to set up reproducible computations (and visualizations) that respond to changes in cells. Cells contain the input data as well as the source code of the computations, and all...
Now that the communion protocol has been ripped, there is no way to get logs, progress or preliminary results from remote transformations. This needs to be tackled in four stages:...
Probably, with buffer_info, the fingertipping and recomputation can be avoided in many cases
Annotate transformations that invoke other transformers as "nesting" in the transformation metadata in the Seamless database. The main use case is to identify all buffers and database entries that are...
Right now, the only resource that is really internal to Seamless is "ncores" (see "multi-core-transformations" test). "memory" should be added as well: - Use bytes2human to parse the user specification...
With the recent update in scratch and fingertipping, it must be made sure that local transformations are indeed still executed locally. A low-level test must be added. In addition, it...
```python ctx.tf = lambda a, b: a + b ctx.tf.a = 4 ``` will give a CacheMissError for "4" This may or may not be a bug. This can be...
This can be disastrous if both processes are compiling the same C code, or building the same modules. Solution: look up tempfile.gettempdir() in the Seamless code and change it.
To indicate the name of the conda environment in which the command is to be executed. Must be known to the assistant/deployment machine(s). Most assistants should be able to deal...
"huge" as being too big to fit comfortably in memory. They are already well supported by bash transformers, because of file zones. For Python, there is currently no alternative for...
Something like `SEAMLESS_SCRATCH_DIR=/scratch` . On the long term, perhaps multiple scratches controlled by `scratch_level`: pyramid of RAMdisk, SSD, rotating disk, etc.