Currently, inputWidth is determined in the following way:
Before measureWrapper is called, it starts out as 400*
Once measureWrapper is called, this.state.inputWidth is set to the resultant this.wrapperWidth
On every onLayoutLastTag call, this.state.inputWidth gets updated to spaceLeft - 10, unless spaceLeft < 100, in which case it gets reset to this.wrapperWidth
* the TextInput width starts out as 400, but its container View starts out with an undefined width
My proposal:
this.state.inputWidth is a function of this.state.text, this.spaceLeft, and this.wrapperWidth
It gets updated whenever one of those inputs changes
It starts out as 90
It gets reset to 90** whenever this.state.text becomes the empty string
Otherwise, the logic is the same as the last bullet point in the existing behavior
** This is meant to be long enough to contain the placeholder. I noticed that without this safeguard, on some platforms (Android I think?) when a new tag is added a newline can temporarily appear and then disappear since the wrapperWidth is temporarily determined based on an out-of-date spaceLeft
Note that this PR is stacked on #37, and is diffed against that PR for easier readability.
Currently,
inputWidthis determined in the following way:measureWrapperis called, it starts out as 400*measureWrapperis called,this.state.inputWidthis set to the resultantthis.wrapperWidthonLayoutLastTagcall,this.state.inputWidthgets updated tospaceLeft - 10, unlessspaceLeft < 100, in which case it gets reset tothis.wrapperWidth* the
TextInputwidth starts out as 400, but its containerViewstarts out with an undefinedwidthMy proposal:
this.state.inputWidthis a function ofthis.state.text,this.spaceLeft, andthis.wrapperWidththis.state.textbecomes the empty string** This is meant to be long enough to contain the placeholder. I noticed that without this safeguard, on some platforms (Android I think?) when a new tag is added a newline can temporarily appear and then disappear since the
wrapperWidthis temporarily determined based on an out-of-datespaceLeftNote that this PR is stacked on #37, and is diffed against that PR for easier readability.