nlmixr2 5.0

nlmixr2 5.0

As time progresses, dependencies update, dependencies come and go, and sometimes these dependencies change what happens to rxode2/nlmixr2 models.

One dependency change (qs being dropped) means that models from before nlmixr2/rxode2 5.0 have the UI/fit models saved to disk will may not be able to load completely in nlmixr2 5.0. This is because internally nlmixr2/rxode2 encoded its models in binary qs format.

Why is qs going away?

While the maintainer wanted to keep qs and its file format around, R is changing and what is acceptable to access within C/C++ from R has changed. This is to try to keep people from misusing these entry points and breaking R.

Since qs had specific code accessing these entry points, this change makes it impossible to have the same file format.

Travers Ching, the maintainer of qs, made a huge effort to create the new qs2 package that uses the R serialization instead of the custom serialization. We are using the qs2 serialization format for cached files and intend to keep that option. (See github post)

We have switched to simple lists and environments for rxode2/nlmixr2 function models. For our C code, we have switched to compressed serialized data, with the option to use qs2 with their superior encode/decode speed. This can be changed with the option(rxode2.serialize.type=""). The following options are currently supported:

• option(rxode2.serialize.type="qs2") – Use the qs2 format

• option(rxode2.serialize.type="qdata") – Use the qdata format (from qs2 package)

• option(rxode2.serialize.type="base") – Use simple, base uncompressed serialization

• option(rxode2.serialize.type="bzip2") – Use base serialization compressed with bzip2

• option(rxode2.serialize.type="xz") – Use base serialization compressed with xz (default)

I would like to thank both the R core team and CRAN for their effort in trying to make R more stable and secure.

I would also like to thank Travers Ching for his diligent effort in making faster binary inputs/outputs from the R ecosystem.

Looking forward

This has highlighted a need for a simple model export method that gives simple text files or a zip file for everything a fit has and possible minimal footprint.

This method will keep in mind the expanded eCTD format and allow you to import or export models.