Re: Totally Gormless

[Armando Di Cianno]

On 2005-10-11 15:59:50 -0400 Nicola Pero 
<nicola.pero@operatelecom.com> wrote:
> > Well ... yes and no.  Sane use of CFLAGS should definitely be up to
> > either the person compiling by hand, or the package maintainer in a
> > distro.  Also, ideally, I don't think that gnustep-make should 
> > compile
> > anything at all, but rather by easily relocatable; everything being
> > based on the env. settings of GNUSTEP_MAKEFILES (only for
> > dev/building) and GNUSTEP_SYSTEM_ROOT.
>
> I'm not sure I get what you're trying to say.
>
> You seem to imply that having the two C command line tools in 
> gnustep-make 
> breaks relocatability.  Can you explain your argument more in detail ?

I suppose "relocatable" was the wrong word.  What I was trying to 
emphasize is that w/o any compiled code, and without "support scripts" 
like GNUstep.sh/csh, the gnustep-make package can be placed anywhere, 
and, conceivably, utilized by just setting the GNUSTEP_MAKEFILES 
variable (maybe with GNUSTEP_SYSTEM_ROOT/Library/Makefiles as 
default).  I feel like this would be the most ideal situation -- one 
that, once I grok what which_lib.c is doing better, I will feel is 
very possible.

> > - - it is only called *once* from GNUstep.sh (currently)
> >
> > [...]
> >
> > ... so it takes about 2 times as long to run -- which still amounts 
> > to
> > a split second. ;-)
>
> PS: That's a lot slower ;-)
>
> .... but looking at your code, I'm certainly not opposed to replacing
> user_home with a sed one-liner if we know that the sed one-liner will
> always work properly (eg, on all platforms) [and I believe I can 
> optimize
> the sed stuff so it goes as fast, if not faster, than user_home.c]. 
> ;-)

Yeah, the posted sed script just keeps chomping on stuff until it gets 
what it wants ... not the prettiest.

> The reason we have user_home.c is that the code to determine the user 
> home
> wasn't exactly simple, and couldn't be done properly with a single 
> unix 
> command ...

This is why I'd like to focus on getting rid of GNUstep.sh/csh -- the 
only place that `user_home` is called.  I've been through 
gnustep-base's NSPathUtilities.m (iirc) many, many times (mainly to 
get my patch of using $HOME if it's defined, as to not pollute /root 
with a GNUstep dir), and I can say that it does a heck of a job 
constructing a workable environment based on the information given in 
GNUstep.conf coupled with "happy defaults" if things aren't specified 
there.

> PS: My original idea was to simply include GNUstep.conf in shell 
> scripts.  I 
> hope the idea has been kept in the sense that the GNUstep.conf syntax 
> is
> compatible with sh syntax.  If so, and if we can just include 
> GNUstep.conf
> in shell scripts (and makefiles!) instead of running C tools, that 
> would
> be as fast and simple as you can get it. :-)

Definitely -- as long as the other variables in that file like 
"SYS_APPS" and all that don't start to pollute a user's environment.  
No reason they would have to, of course, if that file gets source 
while launching apps, not at a user login, like GNUstep.conf.

To illustrate, right now I only have:
GNUSTEP_SYSTEM_ROOT=/usr/GNUstep/System
GNUSTEP_LOCAL_ROOT=/usr/GNUstep/Local
GNUSTEP_NETWORK_ROOT=/usr/GNUstep/Network
USER_GNUSTEP_DIR=
USER_GNUSTEP_DEFAULTS=Defaults
... in my GNUstep.conf, and all of this get sourced by my patched 
GNUstep.sh.  Since gnustep-base defines the "happy defaults" for 
SYS_APPS/etc, I can live without setting them.

> I seem to remember there is lot of hidden complexity in here, but 
> don't
> remember much about the details, so probably reading old email 
> archives and 
> checking what the code does etc. is required! ;-)

I spent an evening with post-its and NSPathUtilities.m when 1.11.0 
make/base were released trying to reconstruct my $HOME-respectng 
patch, trying to flow-diagram the different paths through the code.  I 
feel the only part that makes the code scary is compatibility support 
for GNUsteprc ... Also, I feel like there were many sections of dead 
code because of this; hopefully that will be removed when backwards 
compatible support for GNUsteprc is removed.

> > Honestly, I think a gnustep-make package that was just a bundle of 
> > build
> > files would be much more powerful than the one that gets compiled / 
> > has
> > GNUstep.sh / etc.
>
> I doubt compiling which_lib.c and user_home.c has much impact on 
> this, but
> we certainly all agree about getting rid (or reducing) the need for
> GNUstep.sh ;-)

That's just it!  If user_home is removed, we are pretty much done.  If 
that is removed, then all the settings GNUstep.sh exports can either 
be set in GNUstep.conf, are configure-time options for -base, or rely 
on -base's happy defaults.  Cases like GNUSTEP_FLATTENED probably need 
to be looked at more.

> > I also have patches for gnustep-base, that are path related.  For
> > example, $HOME is ignored in the code that figures out where the
> > Defaults Library gets written, so for e.g., I can specifiy the home
> > for gdomap to not be "/root", which is ridiculous, imho. My main goal
> > is to make them integrate nicely with the current GNUstep.conf based
> > way, rather than make something entirely new, that'll likely be
> > rejected. :-)
>
> Well I'm not sure about that but you should certainly submit your 
> patches
> .... why keeping them hidden ... ;-)

Sure, find them attached.

However, I think it was about last year this time when I submitted 
similar patches against an older -base, and they were rejected, iirc, 
because some were afraid of what would happen if someone used "su" ... 
I however pointed out that "su -" or sudo should be used 'cause (man 1 
su) "The  optional argument  -  may be used to provide an environment 
similiar to what the user would expect had the user logged in 
directly." whereas "su" doesn't guarantee this; i.e. "su" is just 
dangerous. ;-)  Plus, the use of getent to get a user's home dir is 
correct -- I just feel that the root user is a very special case -- / 
or /root shouldn't even be assumed to be writable, imho; plus, a login 
shell gaurantee's that $HOME is correct ... it's up to a user to break 
their own environment. :-)

Another reason for the patch is that I also need it for allowing 
builds to complete while in a sandbox (i.e. Gentoo ebuilds).  The 
simple act of making or installing apps, and docs, would create 
~/GNUstep in /root.

__Armando Di Cianno

<base-user-defaults.patch-1.11.0-3>

<make-user-defaults.patch-1.11.0>

base-user-defaults.patch-1.11.0-3
Description: Binary data

make-user-defaults.patch-1.11.0
Description: Binary data

pgpXCXl3622m7.pgp
Description: PGP signature