R language server reported as available even when language server package is missing (but R is installed)

[krassowski]

While node modules-based specs check for the package presence, the shell-based specs only check for the command existence:

https://github.com/krassowski/jupyterlab-lsp/blob/e393acec2d7d8444ee8ae56c54e644a53b9a308d/py_src/jupyter_lsp/specs/utils.py#L34-L42

In the case of R, Rscript is available whenever R is installed - even if the languageserver package is not installed. We can very easily get a one-liner R script to check if the package is installed. This should probably be a method of SpecBase called in __call__; it could be named is_installed(self) or something like this.

Note: this does not affect python's plys as the checked command (pyls) is directly linked to the LSP server installation.

[krassowski]

The existing code which checks for installation should probably be moved into this new method; the child classes would then be able to call super().is_installed() if they need to access the result of command/node existence check.

[bollwyvl]

Good catch! Is there any way to check it without actually starting a process? Will it be somewhere relative to rscript? Can we do a pattern like the node_modules stuff? I'm pushing towards fewer processes overall, but especially at launch, as lab already fires up a nodejs process in every launch to see if its build is up to date, and with kernels also starting up, this can lead to "too many files open", especially on... wait for it... Windows. But also in little cloud containers, like where we do ci.

…

On Tue, Nov 5, 2019, 15:56 M. Krassowski ***@***.***> wrote: While node modules-based specs check for the package presence, the shell-based specs only check for the command existence: https://github.com/krassowski/jupyterlab-lsp/blob/e393acec2d7d8444ee8ae56c54e644a53b9a308d/py_src/jupyter_lsp/specs/utils.py#L34-L42 In the case of R, Rscript is available whenever R is installed - even if the languageserver package is not installed. We can very easily get a one-liner R script to check if the package is installed. This should probably be a method of SpecBase called in __call__; it could be named is_installed(self) or something like this. Note: this does not affect python's plys as the checked command (pyls) is directly linked to the LSP server installation. — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <https://github.com/krassowski/jupyterlab-lsp/issues/97?email_source=notifications&email_token=AAALCRGY75QFAGISG625443QSHMZBA5CNFSM4JJI7TT2YY3PNVWWK3TUL52HS4DFUVEXG43VMWVGG33NNVSW45C7NFSM4HXAN6KQ>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AAALCRETOSYQOMT5JVSB66LQSHMZBANCNFSM4JJI7TTQ> .

[krassowski]

I guess we could look in library paths, but this would be much less reilable than just using a dedicated mechanism.