bug#30320: 26.0.91; Crash when using lsp-ui-doc-mode

[martin rudalics]

> I considered the case of width, but couldn't find a case where 1 is
> too few.  Is there some configuration where one or more columns are
> taken by some display element other than text?

Is the truncation/continuation glyph mechanism robust enough to handle
one column wide windows?

> Also, cm.c only aborts
> if the Y coordinate is wrong, it doesn't care about the X coordinate.

I have no idea what MagicWrap does (the "Wrap" made me wrongly suspect
a width problem initially).

Maybe a simpler fix would be to process the 'min-width' and
'min-height' frame parameters for graphic frames only.  Then
'frame-windows-min-size' should be able to handle this problem without
further assistance.  Note that 'window-safe-min-width' is 2 which also
gives the minimum width of the frame.  OTOH 'window-safe-min-height'
is 1 and we add 1 for a header line, 1 for a mode line and 1 if the
frame contains a minibuffer/echo area window so we can have a one line
frame if there's none of the latter.

martin