vetr

vetr copied to clipboard

Implement Variable Length Lists with `elist` and `vlist`

1

`elist` (Extensible List, could be `xlist` too) is an extensible list, where objects are accepted assuming that they have every element that is present in the template. This is supposed...

brodieG

Add instructions for utils::globalVariables() for the ., figure out if there is some way the package can inherit that declaration for its checks. Magrittr issue for context: https://github.com/tidyverse/magrittr/issues/29

brodieG

@moodymudskipper: https://moodymudskipper.github.io/typed/

brodieG

Verify Pointer Alignment Compatible with Double

1

We write `(char *)` pointers to a malloc bloc that is `double` aligned; need to make sure the assumption that the alignment is compatible is true, e.g. in startup assertions....

brodieG

@dselivanov asks: > @BrodieGaslam do you think it is feasible to make something similar to https://pydantic-docs.helpmanual.io on top of vetr? Looks like a good idea to signal classed conditions, especially...

brodieG

Reduce Frequency of Garbage Collection

1

Clearly `qassert` triggers way fewer GCs: ``` > gcinfo(TRUE) [1] TRUE > simple_vet nums for(i in 1:5e4) simple_vet(nums) Garbage collection 206 = 155+20+31 (level 0) ... 35.8 Mbytes of cons...

brodieG

In success case, we should not need to match.call the user call. Can it be avoided? More generally, the `match.call` used within functions should be avoidable by using a wrapper...

brodieG

Subset calls not properly recursively substituted

1

``` > val.toks vet(val.toks[[1]], 1L) [1] "`1L` should be type \"language\" (is \"integer\")" ```

brodieG

Extra BMP Unicode and Windows

1

Something is wrong with how windows is handling the out of BMP unicode code points. For example, "\U0001F600" shows up as "" on windows, basically reading the long form U...

brodieG

Add a mode that just returns FALSE on failure instead of constructing an error message in order to have similar performance in success and failure (@franknarf1).

brodieG