I’m strongly against investing time writing such a script, because:
- People may not be able or willing to have a script run over their dependencies. (Imagine telling a middle manager of a financial institution that you have to rely on code that was possibly modified by some “simple script”?)
- Even if we agree to run it, the script must be run manually, thus requiring additional effort on every build of every new module.
- I doubt the script will really be that simple. It took me some time to even understand the issue, and there may be subtle corner cases lurking (e.g. code generated by macros) where your script will fail to match the required code (Having written a few lints for rust-clippy, I can attest that this can happen). Hilarity ensues.
- I disagree that writing such a script will assuage the naysayers or as you say, “send the message that breaking changes will be extremely few and far between” – I think it rather sends the message that as long as the extra steps required to fix it can be automated, however precariously, we are OK with breakage.
- Anyone remembers python’s
2to3tool? The similarity is striking.