Re: [Monotone-devel] Status of nvm.stripped

[Jack Lloyd]

On Mon, Dec 15, 2008 at 12:48:13AM +0100, Thomas Keller wrote:
> 
> Whats the status of this branch? Would people like to see this in
> mainline for 0.42 or is it not yet ready?

I asked Markus about the status a day or two ago, he indicated he felt
it was pretty much working but had not been sufficiently tested yet.

I have been using nvm.stripped as /usr/local/bin/mtn on my desktop
(Gentoo/amd64) for a week or so with no problems. Test suite also
looks OK on this platform.

Building it on FreeBSD 7.0/amd64 shows some snags:

Lua is installed into /usr/local/{include,lib}/lua51 which is not
searched by default. Unfortunately FreeBSD 7.0 does not seem to
include pkg-config info for lua (argh), so this does not get picked up
and autoconf dies as it cannot find Lua. Setting CPPFLAGS, CFLAGS, and
LDFLAGS to search these directories before running configure makes
things happy.

FreeBSD 7.0 packages PCRE 7.4, which is apparently too old. I
installed PCRE 7.8 from source. (FreeBSD 7.1 has PCRE 7.7.something so
this should not be a problem with future releases)

FreeBSD's make is unhappy with the Makefile:

$ make
Error expanding embedded variable

(Yes that is really all it tells me, no line number or anything). GNU
make works fine. I haven't checked if this is specific to nvm.stripped
or not.

Once built:

$ ./mtn version --full
monotone 0.41 (base revision: 0d3abccd30a68d30f70dd35e3f8acf6a49f3698c)
Running on          : FreeBSD 7.0-RELEASE FreeBSD 7.0-RELEASE #0: Sun Feb 24 
10:35:36 UTC 2008     address@hidden:/usr/obj/usr/src/sys/GENERIC amd64
C++ compiler        : GNU C++ version 4.2.1 20070719  [FreeBSD]
C++ standard library: GNU libstdc++ version 20070719
Boost version       : 1_34_1
SQLite version      : 3.4.1 (compiled against 3.4.1)
Lua version         : Lua 5.1
PCRE version        : 7.8 2008-09-05 (compiled against 7.8)
Botan version       : 1.8.0 (compiled against 1.8.0)
Changes since base revision:
format_version "1"

new_manifest [fc676821822bdcba693d9823013d607ce40bd4cd]

old_revision [0d3abccd30a68d30f70dd35e3f8acf6a49f3698c]


With this binary I see one test failure, database_dump_load line 23.
Maybe a SQLite issue?

I think nvm.stripped is the right thing, looking long-term, but Zack's
suggestion of landing it after 0.42 for the widest possible testing to
work out the kinks probably makes more sense than merging it and then
immediately releasing.

-Jack