You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
TL;DR: Desire a documented way for commands intended to be run to generate inputs to themselves depend on inputs
As part of a separate discussion, the seemingly intentional exclusion of task inputs that can be generated by a shell command is very strange to me.
For instance, I cannot use workbench.terminal.run as the command parameter of an input with the type of command.
I am coping with this by installing Task Shell Input, a command by Augusto César Dias (thank you!)
What this allows me to do, is prompt for user input for example, when I want to control or filter which subversion tags are displayed in a user pick list.
When the user selects my SVN - Checkout Blender Precompiled Libs Trunk on Windows, I have to ask them which windows library they want, since Blender Foundation has split up the precompiled libraries based on Visual Studio version.
What isn't working is when the user wants to first select the tag to checkout, and then I want to prompt them with a list of available Windows versions.
Here is the relevant parts of my `task.json` file (click to expand)
I would like to be able to have a task input that would accept a type of shell or process (like task). The user would supply a shell command or process (optionally breaking it down by OS variant [like task]). The output of the shell or process would produce text that would be split by newline in VS Code natively. Each split would be a list item for the input.
What Could Work
If there was some documented way for commands run by inputs to gather inputs themselves, then this would work today.
If you need to collect many inputs, the current recommendation is that you write an extension that has a command that prompts for input. We do not want to make task inputs recursive, since that gets us into the territory of writing a new programming language.
TL;DR: Desire a documented way for commands intended to be run to generate inputs to themselves depend on inputs
As part of a separate discussion, the seemingly intentional exclusion of task inputs that can be generated by a shell command is very strange to me.
For instance, I cannot use
workbench.terminal.run
as thecommand
parameter of aninput
with thetype
ofcommand
.I am coping with this by installing Task Shell Input, a command by Augusto César Dias (thank you!)
What this allows me to do, is prompt for user input for example, when I want to control or filter which
subversion
tags are displayed in a user pick list.When the user selects my
SVN - Checkout Blender Precompiled Libs Trunk
on Windows, I have to ask them which windows library they want, since Blender Foundation has split up the precompiled libraries based on Visual Studio version.What isn't working is when the user wants to first select the tag to checkout, and then I want to prompt them with a list of available Windows versions.
Here is the relevant parts of my `task.json` file (click to expand)
Here is my complete `tasks.json`
What Happens:
What I Would Like (opinion)
I would like to be able to have a task
input
that would accept a type ofshell
orprocess
(liketask
). The user would supply a shell command or process (optionally breaking it down by OS variant [liketask
]). The output of theshell
orprocess
would produce text that would be split by newline in VS Code natively. Each split would be a list item for the input.What Could Work
If there was some documented way for commands run by inputs to gather inputs themselves, then this would work today.
See Also
augustocdias/vscode-shell-command#10
The text was updated successfully, but these errors were encountered: