On Mon, Aug 20, 2001 at 07:43:28AM +0100, Mathieu Olivier wrote:
> I tried to merge QW and QW-server during Sunday (I've not finished but
> that's not the point), and I discovered that most files in the "qw"
> directory have no big differencies (if any) between their alter-egos in
> "qw-server". Which makes me wondering: why don't you try to merge them ?
> (a la ZQuake, maybe ?) The latest the hardest.

I was intending this to be a little surprise, but that will probably not
happen now that it's been suggested and attempted.  These files in
qw-server are not significantly different from those in -qw:

bothdefs.h    crc.h        mdfour.h      pmove.c       spritegn.h
bspfile.h     mathlib.c    modelgen.h    pmove.h       sys.h
cmd.h         mathlib.h    net.h         pmovetst.c    zone.c
common.h      mdfour.c     net_udp.c     protocol.h    zone.h
crc.c

With the notable exception of quakedef.h/qwsvdef.h, these files are
similar enough that they could essentially be deleted from QW without any
effort.  This leaves:

cmd.c         pr_cmds.c    qwsvdef.h     sv_move.c     sys_win.c
common.c      pr_comp.h    server.h      sv_nchan.c    world.c
cvar.c        pr_edict.c   sv_ccmds.c    sv_phys.c     world.h
model.c       pr_exec.c    sv_ents.c     sv_send.c
model.h       progdefs.h   sv_init.c     sv_user.c
net_chan.c    progs.h      sv_main.c     sys_unix.c

So more than 1/3 of the server is already able to be outright taken from
-qw today, which makes sense considering that when I started this project
I intentionally split the server files out of the client to allow a little
bit of natural evolution.  The intent has been to put them back together
at some point all along, hence the reason why I've let things such as a
dependency on SDL happen to the server.


> QW-server is obviously less maintained than QW. Several changes are
> present in QW files and not in QW-server. I even spotted a QW-server
> only bug in Vic recent optimization of the mathlib (mathlib.c:460: '-'
> is missing before zrot[0][1]). This kind of errors could easily be
> avoided by having only one copy of each file at the most.

I noticed this while building the above lists earlier.  Of course, that
function is only used in the client AFAIK, which is why it has not been
noticed in the server.


> Even if I don't really like the "all-DLLs" approch of QuakeForge, I
> think that their (static) libs system is definitely a good thing.

Agreed.  And it could happen that way, we've just been working toward
something we can call a release, damn any real merging for now.  I am now
working on the next stage of merging -qwsv into -qw, the includes.  So far
these files in -qw have been removed from quakedef.h:

cdaudio.h   menu.h
crc.h       sbar.h
cvar.h      sound.h
input.h

A few more will happen soon.

-- 
Joseph Carter <[EMAIL PROTECTED]>                 Free software developer

<Overfiend> partycle: I seriously do need a vacation from this package. 
            I actually had a DREAM about introducing a stupid new bug
            into xbase-preinst last night.  That's a Bad Sign.

PGP signature

Reply via email to