Re: [PATCH v4 7/8] GDB: Add a character string limiting option

["Maciej W. Rozycki" <macro@embecosm.com>]

On Thu, 10 Mar 2022, Tom Tromey wrote:

> Maciej>  NB is there a way to orchestrate a wide character string in a way that 
> Maciej> wouldn't show gibberish on a plain-ASCII terminal
> 
> If the host encoding is ASCII, then I think non-ASCII characters in a
> wide string should be printed as escapes.  This should always happen,
> though see my series about string printing -- fully generic encoding
> support when printing was not implemented for all languages, and that
> series aims to unify it.  That is still pending some updates by me, and
> some missing reviews.

 That sounds good (and I also mean the sources used in the test suite).  

 Also while preparing changes for the docs in v5 I have noticed this old 
NEWS piece:

>>GDB supports wide- and unicode-literals in C, that is, L'x', L"string", 
u'x', u"string", U'x', and U"string" syntax.<<

which I think means we can do verification by hand, because the limit is 
of course also respected for strings supplied from the UI (and even with 
no target connected), e.g.:

(gdb) print "foobar"
$1 = "foobar"
(gdb) set print characters 4
(gdb) print "foobar"
$2 = "foob"...
(gdb)

Though I'd yet have to figure out how to use the syntax quoted in NEWS in 
a meaningful manner, and an artificial program may also be required, 
because:

(gdb) print L"foobar"
No type named wchar_t.
(gdb) print u"foobar"
No type named char16_t.
(gdb) print U"foobar"
No type named char32_t.
(gdb) 

though I guess the data types do not actually have to be system-supplied, 
and it will be enough if they are reasonably defined.

  Maciej