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
Is it possible to runme run --filename test.md in an idempotent manner?
Imagine the first command was a "set something up". The first time, it would run OK. However for subsequent runs, that first command would output errors that I'd like to ignore (because the hypothetical subsequent commands would still be valid).
The text was updated successfully, but these errors were encountered:
There are ways to make the experience better. However, unless you use a devcontainer with all the prerequisites, Runme executing Shell can't know "how valid" the runtime environment is. It's not different from a terminal session.
I can think of several enhancements to improve the status quo. None of which are mutually exclusive.
1. Ignoring errors on the command level
I have used command snippets such as below "successfully" in the past. Essentially, this will turn any non-zero into a zero exit code. It's not ideal because it masks even valid issues. It's a binary choice between erring on false positives or false negatives. Needless to say, this is available now.
2. Toggleable "set -o pipefail" like behavior with a setting
The idea here would be to define, e.g., on the markdown document level, if errors should bail execution or continue on similarly whether all executed cells are conjoined with OR or AND. Runme does not have this setting yet.
It's not inconceivable to define a list of exit codes that are accepted or denied. For example, 127 usually means a command/executable wasn't found. Similarly, it'd be possible to use substrings of error messages to do the same.
Is it possible to
runme run --filename test.md
in an idempotent manner?Imagine the first command was a "set something up". The first time, it would run OK. However for subsequent runs, that first command would output errors that I'd like to ignore (because the hypothetical subsequent commands would still be valid).
The text was updated successfully, but these errors were encountered: