In D28402#714642, @bapt wrote:

What would be the point for it to be an option? I think if it is safe and faster we should just use that code, instead of complexifying diff code with 2 differents code path

Could we make this (mmap) into an option?

Apart from inline comments I would like to ask you to consider adding at least one test case.

Like in another differential, I'm making a style-related note from a quick glance just to subscribe to this review. But I want to test this more thoroughly.

I want to test this soon, but for now I'll just make a style-related note.

Improve --strip-trailing-cr implementation
add a test case
register diff3 tests with mtree

Add --strip-trailing-cr

Include a commit that I omitted by mistake.

In D29493#712586, @pstef wrote:

In D29493#712580, @bdrewery wrote:

By the way it does not build with -DNO_HISTORY.

That's because of D29361, not this one. Will work on it, thank you for the report.

In D29493#712580, @bdrewery wrote:

By the way it does not build with -DNO_HISTORY.

How do I obtain befs.img.bz2? I can see it neither in tree nor in this patch.

This is a truncation that in my opinion should result in program error.
Also note that for a non-chrooted run the paths could already be in the arrays.
So I would do it more or less like below (untested, compiled):

82455a3319a9 corrected the word but the space is still there.

I tested this and it works for me. The patch as a whole will look good to me when synopsis is updated to reflect the change.

Looks good to me.

Consider adding a cross reference to re_format.

This was fixed by 3049d4cc (r330086).

This has been committed as 916e17fd5678.

In D31146#701876, @gbe wrote:

The man page part looks good, but I am a little bit confused, on my FreeBSD systems I see

root 66156 0.0 0.0 12800 1680 - Is 7Jul21 0:04.38 /usr/sbin/cron -s

when executing 'ps axuw'. So it seems, that some kind of a '-s' option is already implemented.

if (fd == -1 || (f = fdopen(fd, "w")) == NULL) and it will LGTM.

I tested the save/restore history capabilities that are built into libedit and so far it looks good to me.

In D26447#664004, @rgrimes wrote:

Hum, not a core system functionality? May I remind you that ftp://ftp.freebsd.org exists? Though it probably does not run the base system ftpd, it DOES run that functionality.

If there are security bugs in ftpd that will force us to release errata. It won't be the vocal users' responsibility to handle it. On the other hand, it wouldn't painful to pkg install ftpd unless you're emotionally attached. The existence of ftpd in base also provokes suggestions to maintain and enhance it (for example with encryption support).

I believe that all reported issues (and more) have been addressed.

Add code to remove duplicates.

Apply some of the fixes suggested by Jilles.

Thanks, I tried to fix all the obvious mistakes. I will have to test and think about the rest of your comments.

Address mistakes pointed out by Alex.

@kevans pointed out to me that my previous patch replaces the name with a dash character, while login(1) only prepends the dash to the original program name, so that's probably what shells expect. I did try to fix that, it seems to work for the few cases that I tested, but in the meantime someone asked about Linux jails, which I looked into. Those can't be properly supported since our getpwnam won't work with the underlying system's structure; so while it is possible to spawn a login shell, the shell would always be /bin/sh which under standard Ubuntu seems to translate into dash, not a great shell for interactive use.

Agree with jamie's comments.

I would also remove the "most interesting" comment.