Subtle bug in soft wrapping?

[sicher]

Hi, I have no insight into how soft wrapping works technically, but I wonder if this is intended behavior. And if it is, if it would make more sense to have helix wrap on whole WORD objects rather than words?

To illustrate, I have the following wrap settings:

text-width = 80
soft-wrap = { enable = true, wrap-at-text-width = true, max-wrap = 25, wrap-indicator = "" }

Now, if I enter the following:

This is a piece of text that will fill more than eighty characters and then wrap,
  expectedly, to the next line.

Here the comma ends up at column 81 (hence why I wonder if this may be a bug?). But moreover, the following space is pushed to the next line. Here I believe it would be nicer if the wrapping would push the WORD "wrap," to the next line:

This is a piece of text that will fill more than eighty characters and then
wrap, expectedly, to the next line.

[pascalkuthe]

I am aware of this. The current width limit is increased by one because people we complaining that lines wrapped one character to early thanks to the EOL character taking a width of 1. I have a more correct fix in #6417 but that will take some time to finish

[sicher]

Ok. Just to clarify: If you change the "wrap," to "wrapx", the "wrapx" will be pushed to the second line.

[pascalkuthe]

Ah sorry I only scanned the issue those are two seperste issues. I was talking about the "wraps at column 81" which is a bug and fixed by the PR I mentioned. Wrapping at WORD doesn't work well for programming languages. There was a tom of discussion about better wrap strategies in the orginal softwrap PR #5420 (but everyone agreed that it should be left to future work). The conclusions was pretty much:

• we probably want language specific wrapling strategies (for prose a streaming version of unicode line breaks) until then word is probably the most widely applicable
• improving the wrapping as described here rework positioning/rendering and enable softwrap/virtual text #5420 (comment)

I don't have time to work on this right now but after the changes to the doc formatter from #6417 land I would be happy o mentor someone to implement that. The code is a bit subtle tough so some experience with rust is needed

[sicher]

Ok, thanks for the clarification! Yeah, the current implementation totally makes sense as the default.

[kj]

@archseer Is this fixed? I just tried this in master with the same settings and the following text (sorry for the silly example):

foo, foo, foo, foo, foo, foo, foo, foo, foo, foo, foo, foo, foo, foo, foo, foo, foo, foo, foo, foo, foo, foo, foo, foo, foo, foo, foo, foo, foo, foo, foo, foo, foo, foo, foo, foo, foo, foo, foo, foo, foo, foo, foo, foo, foo, foo, foo, foo, foo, foo, foo, foo, foo, foo, foo, foo, foo, foo, foo, foo, foo, foo, 

For me that ends up looking like this:

I understood that the idea was to wrap the words onto the next line and not the space?

Edit: Reading #5420 it looks like this behaviour is due to characters like , being treated as a word boundary. But that hasn't changed (in the context of soft wrapping).