REGR: only convert at end for Block.replace_list

[simonjayhawkins]

In Block.replace_list there is some logic to only convert at the end.

before #40082 the convert parameter of Block._replace_regex was honored, but #40082 changed it not convert. This caused a regression, #44864 and was then changed to always convert in #44897. In this PR, we restore the original behavior and honor the convert parameter, which is unused on master.

before #44940, the conversion was not done in Block._replace_coerce for the non-regex case. #44940 changed Block._replace_coerce to dispatch to Block._replace for the non-regex case. Block._replace converts by default.

i've removed _replace_coerce since it appears that the only logic there now is to decide whether to use Block.replace or Block._replace_regex and is only called from Block.replace_list. (The signatures for Block.replace or Block._replace_regex are now consistent in this PR)

I've labelled as a regression but not sure whether we would want to backport this. I have therefore marked as draft and will wait till the reported regressions #45836 and #45601 that are caused by #44940 to be fixed first to avoid merge conflicts with the backport if we don't backport this.

Alternatively we remove the convert logic from Block.replace_list as dead code.

[simonjayhawkins]

this is the convert for the regex case that has been ignored since #40082

[simonjayhawkins]

This is the dispatch to self.replace which converted by default, and didn't have convert parameter.

[simonjayhawkins]

i've removed _replace_coerce since it appears that the only logic there now is to decide whether to use Block.replace or Block._replace_regex and is only called from Block.replace_list. (The signatures for Block.replace or Block._replace_regex are now consistent in this PR)

i've re-instated this since there is also now some logic for None handling there that ultimately we may either want to make consistent for a list-like and scalar to_replace or maybe handle in Block.replace. Either way, that change would not have been suitable for a backport for the fix #46404

_replace_coerce is only called when to_replace is not a scalar.

handling in Block.replace for a list-like (on main) should probably wait till #45725 is resolved.

[simonjayhawkins]

i've re-instated this

oops. forgot to add the convert argument back in. will do that now.

[github-actions]

This pull request is stale because it has been open for thirty days with no activity. Please update and respond to this comment if you're still interested in working on this.