httpd-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Daniel Stone <dan...@sfarc.net>
Subject Re: Apache 2.0.27 and 2.0.28 RPM available
Date Thu, 22 Nov 2001 02:38:30 GMT
On Tue, Nov 20, 2001 at 12:32:07PM -0800, Justin Erenkrantz wrote:
> [ Please inline your patches rather than attaching them unless
>   you are using LookOut!  It makes it awfully hard to review.  ]

What's LookOut! ? BTW, inlining is sorta tricky to do with mutt, and I
haven't yet mastered vim - it was easy with nmh, tho. :)

> On Tue, Nov 20, 2001 at 09:29:19AM +0100, GOMEZ Henri wrote:
> > >>      - Modify apachectl and apxs to use @sbindir@/@progname@ instead
> > >>        of @prefix@/bin/@progname@. A security concert on 
> > >many distrib 
> > >>        (ie Redhat which use /usr/sbin for httpd)
> > >
> > >Check.	
> 
> I've committed the change to use @sbindir@ in apxs.
> 
> apachectl is a much harder beast as we have to get rid of the 
> $(exec_prefix) crap in config_vars.mk (sbindir=$(exec_prefix)/bin - 
> we can't have the ()s in there).  For some reason, I think that is 
> what the Debian folks posted as a sed script.  Daniel?  I don't
> have the time to do this right now.

apache used to have a program called "ubersed", but we don't use that
any more as I think it's a crude hack - patches work just as well.

> For PIDFILE, you meant @runtimedir@, didn't you?  But, it has
> the same problem as @sbindir@.

Yep.

> I also won't change httpd-std.conf as we want them to explicitly
> run apachectl startssl which defines SSL (what you'd really want to 
> change is ssl-std.conf anyway not httpd-std.conf - that check there 
> is right - if you change define SSL to define HAVE_SSL you have the
> right idea).  FWIW, I think it should be enabled if it is compiled
> in, but no one else agrees with me.  =)
> 
> > >>      - update apxs to let user use it's own installed 
> > >libtool instead of
> > >>        the one in $prefix/build.
> > >
> > >Check.
> 
> No.  We want to be picky with our libtool version - the version we 
> built with (i.e. in build/) is the one want to use with 
> apxs-enabled builds.

Why's this?

> > >>      - add a --with-ssl-port as we have --with-port
> > >
> > >Nope, could you please send this to dev@httpd?
> > 
> > Requirement sent yesterday.
> 
> Commented on this yesterday.  I don't like this.

I didn't catch that one.

> > >>      The general goal is to try to have Apache 2.0 more FHS compliant
> > >> without having 
> > >>      to make huge patchs (or perl replace) at each release ;)
> > >
> > >Well, my package manages to fit into Debian FHS, and only has 
> > >11 patches
> > >all up.
> > 
> > FYI, here is a part of what should be patched or change via perl to make
> > apache 2 fit better on FHS from my spec file :
> <snip, snip>
> 
> I have no clue what this is.  =)  Please explain.

A .spec file is what is used to build an RPM. Its format is fairly
self-explanatory, if (IMHO) on some very, very potent crack.

> > I've got some questions for Apache2 developpers :
> > 
> > What prefix represent, a basedir for conf, or binary,
> > or data ?
> 
> I would say binary, but it is all relative to the layout.

My ServerRoot is /etc/apache2, but almost nothing uses that.

> > Should I set $prefix to /etc/httpd2 instead of /usr ?
> 
> I don't think so.

Er, crap. ;)

> > If may fit better with my layout where ALL config,
> > and particulary the one for build subdir live in /etc/httpd2/
> 
> We *should* respect sysconfdir - then all of config stuff goes
> into /etc/httpd2.  I guess you could make a case that build is
> part of the sysconfdir, but I'm not so sure - none of it is
> user-configurable.  It's a different beast.  -- justin

sysconfdir is not respected, as the conf file is defined in configure.in
to be conf/$progname.conf.

:) d

-- 
Daniel Stone						    <daniel@sfarc.net>
<calc> i have a fucked up question to ask... do rubber bands melt at room
temperature for anyone else here?
* doogie_ looks at calc
* liiwi looks at calc
<calc> i am beginning to think my skin is acidic or something
<asuffield> that may answer your real question better

Mime
View raw message