Re: Manual installed apps

We explicitly attempt to ignore user-installed programs, since these 
can cause disastrous conflicts. Although we could attempt to link 
against random libraries on the system, this is a Bad Idea (tm).

- Toby

On 31 Jul 2004, at 14.27, William Kanoff wrote:

> If you install a program manually, can you tell darwinports that the 
> program exists, so it does not try to download and install it again.  
> I was trying to install Sablotron and the port command tried to 
> download and install expat when I already had expat 1.9.8 installed.
>
> Any help would be appreciated.  Thanks.

Re: New additions

http://darwinports.opendarwin.org/docs/

On Jul 29, 2004, at 3:07 AM, Micah Gideon Modell wrote:

> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
>
> I looked on www.darwinports.org and cannot find a page on how to 
> create/configure/start a new port.  I would like to have curlHandle 
> added and I'm willing to do it myself, but I don't know where to begin 
> - - can I get some help?
>
> Cheers!
>
> Micah Gideon Modell
> - --
> http://www.csh.rit.edu/~micah
> http://www.LiveJournal.com/users/micah_gideon
> AIM: Micah Modell
> It would be a sad situation if the wrapper were better than the meat 
> wrapped inside it. (Albert Einstein)
> -----BEGIN PGP SIGNATURE-----
> Version: GnuPG v1.2.4 (Darwin)
>
> iD8DBQFBCMxmst2pCiI+QNwRAh1DAJ4l3ihv494k7Dwrp5URe/6jm5SuwACeMWDx
> Xvb4E1M/uxNRl/fr4EU0CBI=
> =eidt
> -----END PGP SIGNATURE-----
>
> _______________________________________________

Re: Whoops...

On Jul 31, 2004, at 13:08, Jordan Hubbard wrote:

> The port needed updating, that's all.  Sometimes the earliest ports 
> don't stay in sync with other developments in darwinports, in this 
> specific case the mandir update.  I updated it (now at revision 2), so 
> simply update your ports and try again.

Thanks, Jordan.  I updated, and all is well in munchkin land...

j

--
/~\ The ASCII           Justin C. Walker, Curmudgeon-at-Large
\ / Ribbon Campaign
  X  Help cure HTML Email
/ \

Re: Manual installed apps

This is not an adequate answer to the question. I asked this question a 
while ago and also got no joy from the answer.

There are many types of dependencies. If I do, for example, a "port 
install phpMyAdmin", it does not notice that I already have mysql, or 
that I actually have 3 different versions of mysql on my system which I 
can choose between, and so it goes to fetch, build and install mysql, 
with all the other stuff that gets pulled along coming for a ride. Yet, 
phpMyAdmin does not link against mysql. Actually, mysql does not even 
need to be running on the same machine for phpMyAdmin to work. It just 
needs to be running somewhere.

A strict dependency policy is understandable if you are talking about 
linking with libraries. But the darwinports architecture does nothing 
to capture other sorts of dependencies. Not all things that are 
dependent on each other are dependent in the same way.

There needs to be a way to let a user say, "that dependency is 
satisfied by this" and have that override. I should be able to install 
phpMyAdmin and point it to ~mysql and I should be able to say: "there 
it is!", and "do not re-install this!".

Having different kinds of dependencies raises all kinds of questions, 
and I don't have all the answers.

But you all make it sound as though the question has been asked and 
answered. It has not.

The fact that you have cast the semantics of dependencies into such 

Re: Manual installed apps

Ray,

OK, here's the deal.  Have we "answered" this question?  No, we have 
not answered it any more than a man who's walked to the edge of a high 
precipice and looked down has figured out how to safely descend without 
the aid of wings, rope or parachute.   About all we know right now is 
what *doesn't* work:

A) Simple checks for "foo" that allow other things which depend on foo 
to assume it's there for all potential values of "there" and thus use 
it safely and without inexplicable side-effects that users report but 
developers cannot reproduce.   Short of devising a system for doing 
arbitrarily complex checks for foo that can actually determine 
functional equivalency with close to 100% accuracy, and even with such 
a system it's left as an exercise to the reader just how to code up 
those checks, about all we can realistically do is force the user to go 
down the same path that the developer did when they wrote the port and 
got it to go all the way through its functional tests.   Yes, that 
means installing other components like mysql "gratuitously", but it 
also means that we're not fooled by half-baked or semi-crippled 
installations of mysql that someone else might have dumped onto the 
system.  Apple itself is guilty of this when it comes to shipping 
libraries and headers that are in sync and remain in sync across 
software updates or decoupled updates to the system and the DevTools.  
Sometimes the only way a given port CAN work is to install its 
dependencies itself, even though a quick and naive inspection of the 
system would tend to suggest that a given dependency is already 
"there".  These are not merely hypothetical scenarios, we've seen this 
born out through actual experience and user bug reports.  It's enough 
to drive a porter nuts.

mmosaic?

Hi,
in FreeBSD there is a port, mmosaic, that's very like the original 
Mosaic web browser.
Noone has ported it still in Darwin Ports?

regards

Salvo Micciché

_______________________________________________
Darwinports mailing list
Darwinports <at> opendarwin.org
http://www.opendarwin.org/mailman/listinfo/darwinports

Status of cvsup1.opendarwin.org

I've been wanting to access the opendarwin cvsup server,  but it
continuously refuses the connection.  What is its status?

Greg Shenaut

XFree86 install fails with fonts.conf gripe

I've tried installing both XFree86 4.3.0_2 and 4.4, and in both instances
I get the error:

  Error: Target com.apple.activate returned: Image error: /private/etc/fonts/fonts.conf already
exists and does not belong to a registered port.  Unable to activate port XFree86.

This is under Darwin 7.4.  It looks like the fonts.conf was part of the
original system.  Any clues?

thnx,
mb

Odd dependency problem

I've been trying to get a completely clean DP installation here on Panther,
and it is the first time I've used the new port images stuff.

I've hit what feels like the same glitch now a couple of times while
building all the bits of gnome that I need. Unfortunately after tooling
around a bit the problem seems to go away, so this isn't a bug report, more
a sanity check 

I think the scenario is that port A depends on port B which depends on port
C. Port B is not installed but port C is. Port A fails to install because it
tries to install C anyway and this fails.

eg:

% cd gnome/libgnomeprint
% sudo port -f install
--->  Installing librsvg 2.6.5_1
Error: Target com.apple.install returned: Registry error: librsvg 2.6.5_1
already registered as installed.  Please uninstall it first.
Error: The following dependencies failed to build: eel librsvg
% cat Portfile
# $Id: Portfile,v 1.10 2004/06/08 09:58:42 olegb Exp $
PortSystem 1.0
name            libgnomeprint
version         2.6.1
description     This is an implementation of the Gnome Printing Architecture
long_description        This is an implementation of the Gnome Printing \
                        Architecture
maintainers     gnome-darwinports <at> opendarwin.org
categories      gnome

Re: Manual installed apps

A very cogent response. I would suggest, actually, that Jordan's note 
should be in the 'Darwinports Guide' somewhere. The fact that we would 
acknowledge where we are in the development of this system, and that we 
have not gotten to the point where we can address certain kinds of 
concerns, would help people get a sense of what darwinports does do. 
Developers and users would spend less time being frustrated and 
confused about what it does not do.

And since Jordan threw down the gauntlet, I will describe what I 
believe could be done. I would like to find a way to help with this 
effort, as opposed to just talking about it. I do have to get over some 
barriers first, such as not really doing Tcl, having to do my day job, 
trying to see my kids, and such as that. But here it is.

Various pieces of the puzzle:

1) a port identity/verification system
2) an automated testing infrastructure
3) a rule-based dependency declaration system
4) automated dependency rule authoring

Port Identity/Verification System:

There needs to be a way to actually be able to determine: "yes, the 
port of xyzzy is actually on this system and it is actually complete". 
For example, I have an extension to WebObjects that I wrote. (Sorry to 
bring this up, but WO is my real job and I can either just describe 
things abstractly or use WO examples. This does not seem a stretch as 
so many things that were first in WebObjects are now appearing in Max