What is the problem this feature will solve?
I have a wrapper executable around shfmt that requires multiple arguments to be passed to it, e.g. /path/to/formatter --shfmt. To my understanding, the current LSP setting only accepts a string and not a list of strings for running the formatter, so the command will probably fail with Command '/path/to/formatter --shfmt' not found or something like that.
What is the feature you are proposing to solve the problem?
It would be nice if the LSP setting accepted a list of strings (with the current string type getting "promoted" to a singleton list), so I could have my setting be: [ "/path/to/formatter", "--shfmt" ]. Then in the language server, -- could be appended first, before all of the other shfmt-specific flags that are controlled by the other LSP settings.
I haven't tested this, but I imagine if the user provided -- themselves, then maybe they could provide their own set of shfmt flags themselves? And so the language server wouldn't need to create a new option for every format option that exists.
What alternatives have you considered?
Running a regular version of shfmt, but the version may differ from what's in my wrapper script.
What is the problem this feature will solve?
I have a wrapper executable around
shfmtthat requires multiple arguments to be passed to it, e.g./path/to/formatter --shfmt. To my understanding, the current LSP setting only accepts a string and not a list of strings for running the formatter, so the command will probably fail withCommand '/path/to/formatter --shfmt' not foundor something like that.What is the feature you are proposing to solve the problem?
It would be nice if the LSP setting accepted a list of strings (with the current string type getting "promoted" to a singleton list), so I could have my setting be:
[ "/path/to/formatter", "--shfmt" ]. Then in the language server,--could be appended first, before all of the othershfmt-specific flags that are controlled by the other LSP settings.I haven't tested this, but I imagine if the user provided
--themselves, then maybe they could provide their own set ofshfmtflags themselves? And so the language server wouldn't need to create a new option for every format option that exists.What alternatives have you considered?
Running a regular version of
shfmt, but the version may differ from what's in my wrapper script.