And don't bother running configure and make upstairs; here is enough.

The way to update nano.pot is to run 'make nano.pot-update' in the po
subdir.  Updating a specific PO file is then done with 'make xx.po'.

(The script mistakenly included all leading comments, not just the
ones marked with "TRANSLATORS:".)

It sounded like an instigation to upgrade, or like a list of
instructions on how to upgrade.  But it is merely a list of
noteworthy changes -- what other packages put in the NEWS file.

This fixes https://savannah.gnu.org/bugs/?48246.

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>

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

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.