parentheses highlighting inconsistent on closing parentheses

[darkfeline]

show-smartparens-mode behavior is inconsistent on closing parens.

On opening parens:

*    *
((()))
^

When the cursor is at the caret, the two starred parens are highlighted, but for closing parens it does this:

 *  *
((()))
     ^

The fix is to set sp-show-pair-from-inside to t:

*    *
((()))
     ^

[TheBB]

Maybe you can PR this?

[sooheon]

This was a conscious decision, related to the change with evil-move-beyond-eol (439278b) . I think syl20bnr's decision was to simply adhere a bit closer to the emacs idea of cursor location, which is important for evaluating sexps.

While it may be unnatural when you move between parens by %, it is completely natural and correct if you use paredit/smartparens/lispy commands to move by sexps. In other words, it's not really a bug, but there is room for personal taste and I think a documented toggle may be needed.

[darkfeline]

Derp, this is basically a variant on the other issue I made: #2525

I don't know what, if anything, should be done for this. Maybe spacemacs needs a toggle that sets up everything properly depending on whether evil-move-beyond-eol behavior is wanted or not?

[sooheon]

I think this paren highlighting behaviour is already dependant on evil-move-beyond-eol, so theoretically simply setting that to nil should take care of both issues.

[TheBB]

The user still has to set both, because there's no user code that runs between (setq evil-move-beyond-eol nil) in init-evil and (setq sp-show-pair-from-inside (not evil-move-beyond-eol)) in init-smartparens.

@syl20bnr Since additional packages have dotfile as their owner, maybe we could support things like dotfile/post-init-xxx? I admit I don't have a ton of use cases for such a feature, but maybe it wouldn't be so hard to get it to work, if the dotfile is already "almost a layer."

[syl20bnr]

@TheBB I believe that dotfile/post-init-xxx should work out of the box :-)

What I would like in Spacemacs is the alternative fix: the parens behind the cursor block should be always highlighted. It is better for dealing with lisps and I feel that sane natural defaults in Spacemacs should go towards lisps as Emacs is the host.
For now this is a mix of the two approaches and it sucks...

We should have a setting in Spacemacs to choose to highlight current or previous pair, with a default setup to previous.

It may require some PR for smartparens/evil (I don't know exactly where is the inconsistency - if there is one).

[syl20bnr]

For the evil-move-beyond-eol issue, I think we can get best of both worlds, I'll look into evil to propose a PR to have consistent $ and x even with evil-move-beyond-eol set to t.

[TheBB]

@TheBB I believe that dotfile/post-init-xxx should work out of the box :-)

I confess I didn't actually try, but that's not what I see. It's off topic for this issue though, so I can look at it later.

[syl20bnr]

(@TheBB indeed, doftile owner is a special case: https://github.com/syl20bnr/spacemacs/blob/develop/core/core-configuration-layer.el#L723-L726 should be trivial to add support for an optional init function)

[syl20bnr]

This is a pretty old report, @darkfeline do you still have this behavior ?

[darkfeline]

Nope, it's fixed for me.