two problems encountered trying to add newline to "expand..." text

[brianharvey]

So, sathvikrias posted this improved IF reporter in the forum:

I really like that, but if you have more than one ELSE IF the block gets very wide very quickly, so I wanted to put a line break before the words "else if" so I tried "$nl else if" and got this:

That's the first problem; $symbol names should end at a space and be mixable with other text. The second problem is that, having read about some feature involving "%nl" I also tried "%nl else if." Having done that, I am unable to click OK on the block editor. No error message, it just silently doesn't close (but it does change the block, badly). Even if I delete the "%nl" in the "expand..." option, that doesn't let me click OK. All I can do is click Cancel, which leaves me with the messed up version of the block.

[brianharvey]

P.S. I think he's right that (after fixing the nl problem) we should change the primitive to do this.

[jmoenig]

that's right, newlines and symbols currently cannot be mixed with other label elements in the expansion tokens. Known (because I didn't yet need it, and on my list.
I don't want to add a variadic if/else reporter just yet, I'm not convinced... but let's discuss this sometime offline (in a few weeks). Also he probably shouldn't be using RESHAPE...

[brianharvey]

P.P.S. I did more playing around with how the block is line wrapped and got this:

That's pretty hideous, imho. I get that you're treating the entire multi-line input group as a single token for purposes of wrapping the overall expression, but it took quite a while for me to see that the ELSE floating by itself at the right edge isn't vertically aligned with the text starting THEN.

I agree in principle that it's good to keep the input group together, but if so, it should be followed by a newline automatically. And also if it needs to be broken, do it at a group boundary (so the line starting THEN would instead start with the ELSE IF that comes before it).

Happy to discuss later, but why not RESHAPE?

[jmoenig]

the solution to this is going to be letting block authors insert newlines where they wish to.
What's the point of RESHAPE here, what does it do, and why? Did you try leaving it out?