Control indentation of nested projects by UIManager properties

[jtulach]

• Show nested projects co-located and indented #9602 introduced visualization of nesting
• however it hardcoded the behavior and its L&F
• this PR adds two properties read via UIManager (one new and one shared with other modules) to allow L&F customization or ...
• even compete disabling of the nesting support
• the PR follows the plan acknowledged yesterday

[neilcsmith-net]

Thanks! Looks good on reading. Not tested yet. Probably needs squashing before merge.

Handles my concerns about customization / disabling the display. I still think the node prefixing approach is flawed against filtering the hierarchy, but if everyone else is on the board with the current approach then this is good to go.

Please don't merge this until @mbien has followed up here or on the sorting review though, just in case that gets complicated in master.

[mbien]

tested:

# no indentation (use quotes otherwise it is null)
nb.project.indentation=""

# friendship mode worked too
nb.project.indentation="♥ "

while testing I was wondering if we should split it into three properties: prefix, repeat and postfix?

e.g a user could customize it like:

added bonus is that even without using a prefix, this can produce very compact indentation since there is no need to add whitespace to the arrow. The whitespace can be in the postfix.

nb.project.indentation.prefix=""
nb.project.indentation.repeat="»"
nb.project.indentation.postfix=" "

pipleline symbols:

nb.project.indentation.prefix="└"
nb.project.indentation.repeat="─"
nb.project.indentation.postfix=" "

dark theme / expanded:

dots with space in repeat (just like with the arrow now):

nb.project.indentation.prefix=""
nb.project.indentation.repeat="• "
nb.project.indentation.postfix=""

pacman would work too but my font doesn't support that symbol :(

code ( + formatted using 4 spaces):

Details

    static String indentationPrefix(boolean html, int depth) {
        String prefix = UIManager.getString("nb.project.indentation.prefix"); // NOI18N
        if (prefix == null) {
            prefix = "";
        }
        String repeat = UIManager.getString("nb.project.indentation.repeat"); // NOI18N
        if (repeat == null) {
            repeat = "\u00bb";
        }
        String postfix = UIManager.getString("nb.project.indentation.postfix"); // NOI18N
        if (postfix == null) {
            postfix = " ";
        }
        if (depth <= 0 || (prefix.isEmpty() && repeat.isEmpty() && postfix.isEmpty())) {
            return "";
        }
        if (html) {
            try {
                String xml = XMLUtil.toElementContent(prefix + repeat.repeat(depth) + postfix);
                return "<font color='!controlShadow'>" + xml + "</font>";
            } catch (CharConversionException ex) {
                throw new IllegalStateException(ex);
            }
        } else {
            return prefix + repeat.repeat(depth) + postfix;
        }
    }

just an idea - is this too much flexibility?

[neilcsmith-net]

just an idea - is this too much flexibility?

I'm tempted to say, yes too much, except that pipeline example does look quite nice!

I'm curious where the null is coming from for the empty String. FlatLaf issue?

[mbien]

I'm curious where the null is coming from for the empty String. FlatLaf issue?

likely. Something seems to equate "" with null during parsing. Plain old properties don't do that usually.

I'm tempted to say, yes too much, except that pipeline example does look quite nice!

yeah I like that one too. The reason I experimented with this is because it looked like the symbol+space pattern appears to be fairly limiting in what symbols you can use. Sometimes you want to repeat a symbol and postfix a space, other times you want to repeat the space and add a symbol at the end - or use different symbols like in that pipeline.

[neilcsmith-net]

Something seems to equate "" with null during parsing. Plain old properties don't do that usually.

Yes, I know. I was surprised reading that as I thought that was how we'd tell people to switch it off. Still, as long as we know, we can explain.

At a guess it's due to the code at https://github.com/JFormDesigner/FlatLaf/blob/e735c4898e6772c734229c688976faceab1bdfc8/flatlaf-core/src/main/java/com/formdev/flatlaf/UIDefaultsLoader.java#L461 Makes sense given the values are typed I suppose, or maybe intended to allow override removal.

[jtulach]

Yeah, the pipeline symbol looks pretty nice:

nb.project.indentation.prefix="└"
nb.project.indentation.repeat="─"
nb.project.indentation.postfix=" "

I don't mind changing the algorithm to use prefix, repeat and postfix UI manager keys. Done in bb682d9

[mbien]

code looks good to me and it works. Maybe wait a little more before integration since the impl changed after PR was already approved to give a chance for replies.

(+ don't forget to squash)

[jtulach]

code looks good to me and it works.

• yeah, it should work
  • it is your code (slightly modified)
  • there are tests
• but real verification is always nice

... wait ... before integration since the impl

Yeah, that's a good idea. Re-requested @neilcsmith-net and @MartinBalin reviews.

(+ don't forget to squash)

I'll do my best to not forget about it. Or should I optimistically pre-squash before getting the new round of reviews? Squashed into 5b05d8f

[mbien]

Or should I optimistically pre-squash before getting the new round of reviews?

sure, feel free to do that. There is also the pre-integration checklist on the default PR template with 4 checkboxes to press before merge but few are using it (which is fine).

[jtulach]

Squashed. Ready to merge. Can I merge, @neilcsmith-net?

[neilcsmith-net]

Still looks good to me. Bit more complex, but @mbien example showed the value of it. Slight nitpick on the docs.

[neilcsmith-net]

Nit : there was one, then three, never two! 😄

[jtulach]

Good catch. Using "these" since 79df714