Magnus Ulimoen

There are unfortunately some drawbacks to making the static builds standard, * Increased compile time (primarily on first build) * Licensing (although this could work on an opt out basis,...

@aldanor That is some really great stuff! So it sort of acts as an inplace conversion? Nasty trick of copying the layout of the String :+1:

I believe [nc-config.cmake.in#76](https://github.com/DennisHeimbigner/netcdf-c/blob/d594f72a6d5449cdf5ceaca4050f4e277fad5132/nc-config.cmake.in#L76) should also use `HAVE_H5_SZIP` instead of `USE_SZIP` and the szip feature should be exposed by `netCDFConfig.cmake.in`

Over on the Rust wrapper (https://github.com/mhiley/rust-netcdf) we are wondering how unsafe the library is when multi-threading. * Could we assume reading from the same file from two places simultaneously is...

When working on the wrapper for netcdf, we found it was easy to corrupt internal data structures when working with a non-threadsafe `hdf5` install. In essence, every call to `netcdf`...

We are also experiencing the same problem with reading from sqlite in a loop leading to ```console pyo3_runtime.PanicException: called `Result::unwrap()` on an `Err` value: Os { code: 11, kind: WouldBlock,...

I think this might be due to `CURLOPT_NETRC` being set to the default `CURL_NETRC_IGNORED`. If specifying `HTTP.NETRC` through `netcdf_rc_set` it works as it is supposed to.

The standard netcdf library on Ubuntu machines do not come with parallel support, so this is not supported as of now. The global lock acts to serialize all access to...

The underlying netcdf library does not support accessing the raw bytes, so this is not exposed here either.

Parallell reads/writes goes through pnetcdf. I think this issue can focus on strictly the parallell functionality, which requires opening through `nc_{create/open}_par`