To add a character, one does not need to allocate twice its size.
And the amount to be moved does not depend on the size of the new
character; it just needs to include the terminating zero.

(This fixes in do_output() the same logical mistakes that were fixed
in do_statusbar_output() last February, in acf19bde and 7c0e4333.)

Because when undoing a paste at the end of the buffer, the tail line
probably does not exist any more.

This fixes https://savannah.gnu.org/bugs/?48222.

(It would have been the proper fix for
https://savannah.gnu.org/bugs/?44488.)

Signed-off-by: Benno Schulenberg <bensberg@justemail.net>

This fixes https://savannah.gnu.org/bugs/?48007

.
Tested-by: Rishabh Dave <rishabhddave@gmail.com>
Signed-off-by: Benno Schulenberg <bensberg@justemail.net>

Specifying an operating directory should either lead to a successfull
confinement, or nano should fail to start.

(Also: save the terminal settings as soon as possible, so that an early
die() will not restore uninitialized values.)

This fixes the first part of https://savannah.gnu.org/bugs/?47798 properly.

This fixes https://savannah.gnu.org/bugs/?48103.

(The fix is wasteful -- it should only discard the multidata if actually
the name *did* change, *and* if the applicable syntax changed.)

The mark has already been unset by do_cut_text().  And any assignment
to 'mark_begin' is useless when 'mark_set' is FALSE.

This fixes https://savannah.gnu.org/bugs/?48108,
and fixes https://savannah.gnu.org/bugs/?48157.

Implement its single use case by making better use of another element.

By moving the test for the only situation where do_comment() does nothing
(when only the magic line is selected) further up front, the add_undo()
can be called before the commenting/uncommenting starts and two variables
and an extra function are made unnecessary.

This fixes https://savannah.gnu.org/bugs/?48156.

Reported-by: Mike Scalora <mike@scalora.org>

Some platforms don't like that.

This fixes https://savannah.gnu.org/bugs/?48099 reported by Mike Scalora.

Don't catch the NULL earlier, because we do want an error message when
opening the lockfile fails.

That is: don't mix the number of lines read with a warning about the
file being unwritable -- the former is just convenience information,
the latter is a must-see.

This fixes https://savannah.gnu.org/bugs/?48047.

When running in a non-UTF locale, and when strncasecmp() suffers from
the same defect as strncmp(), make sure not to pass a length with the
high bit set.

This reverts commit df8c3deb from six years ago, which prevented a crash
on the Armel/Maemo platforms but causes nano to lose history items.

The strncmp() function on those platforms treats size_t numbers with
the high bit set as if they were zero.  To avoid that, use a number
that cannot be seen as negative, as suggested by <alpha@qzx.com>.

This fixes https://savannah.gnu.org/bugs/?48048

.
Tested-by: Tito Ragusa <farmatito@tiscali.it>
Signed-off-by: Benno Schulenberg <bensberg@justemail.net>

The function is_cntrl_mbchar() has always been called successfully before
calling control_mbrep(), so the passed character *is* a valid sequence.

Store the file sizes from before and after the commenting/uncommenting
in the undo struct, so they can be restored when undoing or redoing.

This fixes https://savannah.gnu.org/bugs/?48062

.
Signed-off-by: Mike Scalora <mike@scalora.org>

When opening a file that was edited before, and the remembered position
is near the end of the file, then don't center the target line but show
the last line of the file on the bottom line of the screen, thus showing
as much of the file content as possible.

This addresses https://savannah.gnu.org/bugs/?46243.

Instead of parsing every multibyte character twice, first with
parse_mbchar() and then with mbtowc(), just let mbtowc() do all
the work.  This makes searching for a fixed string twice as fast.

This also gets rid of four variables and lots of memory allocations.
(And, more importantly: it stops nano messing up the internal state
of the multibyte-to-wide character conversion, and thus would make
the calls to mbtowc_reset() superfluous.)