Don't remove trailing whitespace about wrap-ansi

[SamVerschueren]

I think we have an issue in log-update now as well (reported here sindresorhus/log-update#20). We started using wrap-ansi but because it trims all the whitespaces, some of the use cases are now foobar. I started to look into this but couldn't come up with a good solution yet. Maybe I should just do the PR (with what I have) and we can iterate from there?

// @sindresorhus

[dthree]

@bcoe we should discuss this, as adding trailing whitespace might break other implementations.

For example, take the string a b c d e, wrapped on 3, and for clarity we'll make * represent white-space. We can do one of two things:

// 1. Allow overflow for white-space only
a*b*
c*d*
e

// 2. Bump the beginning of the next line with white-space:
a*b
*c*
d*e

2 above is uglier, but it won't break anything. 1 above would seem to break things as it violates the expected width of the wrap.

What do you think?

[bcoe]

@dthree my gut would be that we should opt for option .2; especially when hard is enabled, we can never afford to overflow since tools like cliui and yargs use wrap-ansi for multi-column layout (columns would start to flow into each other).

Either approach taken, it's probably best to opt for a major bump I think?

[SBoudrias]

Maybe we could just keep the whitespace at the beginning and the end of the block?

[bcoe]

@SBoudrias this sounds reasonable 👍 if you have the time, could you send a failing test case perhaps?

I'd love to see Inquirer using this, big fan 👍

[dthree]

@bcoe I agree on both points.

@SBoudrias yeah - could you please give some examples to ensure we understand you correctly?

[sindresorhus]

and we can iterate from there?

👍

[axe312ger]

Yes please make that configurable! <3 :)

[axe312ger]

thank you! 🙇