The current behaviour of storing the scratch file in
the current directory is a bit painful, because it
generates files in all the directories where you
execute ed. BSD ed uses TMPDIR for this purpouse,
so if the user wants to put the scratch file in
other place different of /tmp it only has to set
this variable.
This patch introduced init() function, which removed the initialization
code of savfname in doread, but this is incorrect, because savfname
can be initialized with a r command if savfname is empty.
These funcction initializes the scratch buffer, and in case of
having a file name parameter it loads the file in the scratch.
It also fixes a problem before this version, where the saved
filename was not set when the file didn't exist.
For the same reason we don't care about the state of
the buffer in clearundo() we can reset modflag in
clearbuf(), because after calling it we don't want
to know anything about the previous state.
When we discard the content of an undo buffer
is because we don't need it anymore, and it means
that we don't care about the modify state of the buffer
so we can reset the modflag in clearundo(), and
remove this assignation each time clearundo() is called.
When we move the exit() out of venprintf(), we can reuse it for
weprintf(), which basically had duplicate code.
I also renamed venprintf() to xvprintf (extended vprintf) so it's
more obvious what it actually does.
getrhs() must remove \ excepts in the case of & and \d
(where d is a digit), because in this case are sequences
understood by addsub(), so addsub() must be able to see
them.
The majority of the systems define PRIO_MAX and PRIO_MIN,
but there is an obscure system, whose name I am not going
to tell, where they were not defined.
Every system is going to have a different configuration
so the only solution is to put an ifdef guard for every
value. To do this, we generate the header at compile time
with a shell script.
When a error occurs it is important to remove all the modifications
done by the offending command and restore the value of dot to the
state before of executing the command.
by re-ordering when chmod/chown is done, only a list of directories (not
all files) need be kept for fixing mtime.
this also fixes an issue where set-user-id files in a tar may not work. chmod
is done before chown and before the file is written. if ownership changes, or
the file is being written as a normal user, the setuid bit would be cleared.
also fixes ownership of symbolic links. previously a chown() was called,
which would change the ownership of the link target. lchown() is now
used for symbolic links.
renamed all ent, ent* functions to dir* as it better describes what they
do.
use timespec/utimensat instead of timeval/utimes to get AT_SYMLINK_NOFOLLOW
This reverts commit a564a67c4ea70e90a4dc543814458e4903869d3e.
Not as trivial as I thought. This breaks cp when used as:
cp -r /foo/bar /baz
The old code expands this to:
cp -r /foo/bar /baz/bar
This can happen if you move a file from one filesystem to another.
rename(2) will fail and we will fall through to a manual cp + rm.
Initiate the rm through recurse() like we do for rm(1).
Thanks to Heiko for reporting this.
