Re: GNUstep ready to go! (was: The state of GNUstep packaging)

[Michael Baehr]

On Sat, 5 Jun 2004 13:43:27 +0200, Andreas Höschler
<ahoesch@smartsoft.de> wrote:

> The "tight-knit community"!? Just a few thoughts of mine. I get the
> impression - response rate on the list is very high in the evening and
> on the weekend, less high at normal business hours - that GNUstep
> currently is a toy for some freaks that learned to appreciate the
> advantages of Objective-C and the open source idea behind the project.
> That's ok for me. I consider me being such a freak. However, it at
> least  seems to me that GNUstep has less momentum in the market
> (Software Industry) than it could have. You can do (at least) two
> things with GNUstep from a commercial point of view (hey, we all need
> to do something for our living, why not use GNUstep for this as well).

GNUstep is, indeed, the domain of hackers as far as a lot of the
development goes (and especially the use of the more desktop-oriented
applications).  From what I understand, however, there are a
reasonable number of corporate (and other "serious") users who are
deploying GNUstep on the back end (in many cases as a migration path
for their apps away from the abandoned OPENSTEP platform), but we
don't seem to hear from them too much.   I will refer to this again
further down.

> 1) You can develop a super fancy GNUstep-based application that
> everybody wants to have, e.g. a scientifical workdprocessor that blows
> "Scientific Word" away and hope to sell it 1000.000 times to compensate
> your development costs. Here we have the problem of the tiny
> installation base. The goal of the "GNUstep packaging" project as I
> understand it is to have a ready to install (with a double click)
> GNUstep package on any Linux distribution CD, so that Linux
> installation base = GNUstep installation base. Then providers of above
> applications would have a decent target platform. I pretty much
> appreciate the efforts in this direction. I bet as soon as this target
> platform (installation base) is there, vendors will be more willing to
> invest money in portig or redeveloping apps on GNUstep. We (Smartsoft)
> at least would...

That's exactly my goal, and I believe my fellow packagers share it :) 
We want as many platforms as possible to be GNUstep platforms as well.
 A "killer app" could well suffer and enjoy a lukewarm response if too
many people are unwilling to go through the trouble of installing
GNUstep due to nonexistant or outdated packaging.  At the same time,
it could also inspire a lot more distro/OS attention to GNUstep as a
whole (if the app was particularly "killer") ;)

> 2) You are in the project business. You develop highly customized
> business applications and are also responsible for the installation and
> maintenance at the customer. Here a GNUstep installation package is no
> issue. You can always install GNustep from CVS since you (the hacker)
> are the one that installs everything, not the user of the software
> (customer). However, there are still other issues to be aware of. Open
> Source projects are evolving and per se can't be as stable as e.g.
> MacOSX/Cocoa (more testing before each release). We were more than once
> confronted with the fact that we had something running with one version
> of GNUstep, installed a newer one (because a newly implemented feature
> was needed) and our application broke, since other stuff in the
> meanwhile had been modified that caused severe problems with our app
> and/or environment. This should not be an attack against GNUstep, the
> opposite is my aim. This simply is the nature of open source projects
> and one has to live with that and spend enough time on testing and
> fixing a new GNUstep version before putting into production.

As I mentioned up above, I believe there are a significant (at least
in GNUstep terms) amount of people in this exact position, and, yes,
updating the frameworks (and their applications) can currently pose
quite a challenge on many platforms.  Proper packaging will solve this
to some extent, but the issue of maintaining API coherency across
versions remains.  I feel that the GNUstep API has pretty much
stabilized at this point, however, especially if you look at what has
been implemented and the current version numbers (which I feel
accurately reflect the current status of the project).  Hopefully this
will become less and less of an issue in the months to come.

> I see that there is work in progress to improve problem 1). I was
> wondering what we could do to help with challenge 2). Customers are
> happy if they get powerful applications in a short time for a payable
> amount of money. Cocoa/GNUstep/OPENSTEP should always be the choice
> here. We had some decent success in convincing customers to set on this
> platform. But I think here is much more potential. It would help a lot
> if we could make the power of GNUstep and the community of GNUstep
> developers more transparent to potential customers.

Before addressing your suggestion below, I'd like to mention that I
think one of the most significant barriers here is not only a lack of
information but the fact that deploying an application on GNUstep
first is currently a bit of an uphill battle.  For documentation, we
generally rely on Apple's docs.  The automated tools that made
OPENSTEP such a simple environment to produce software for are either
missing or lacking in features on the GNUstep side (with the exception
of Gorm, which I feel is at this point a 1-to-1 match for Interface
Builder in terms of usability).  Because of this, I'd say the majority
of serious users of GNUstep are probably those with existing OPENSTEP
or OS X code bases that they are migrating, where the majority of the
work is not in rewriting code or interfaces (which have already been
created), but in the minor tasks of adapting it to build and work on
GNUstep.  I'd like to see this change and GNUstep become more and more
of a first-time-deployment platform.

> I for example would like to see a maintained list of
> 
> • professional Cocoa/GNUstep developers (name, location, contact
> parameter, availability,...)
> • companies that are specialized in developing application for
> Cocoa/GNUstep in Objective-C
> • Job Offerings
> • Projects (GNUstep based) / Features (still missing in GNUstep)
> currently being developed by whom
> 

Yes.

> With projects I don't mean "Spare time developer XX is working on
> ImageViewer.app" or the like. With projects I mean "Company YY is
> working on a 200.000 EUR project (GNUstep based) for customer ZZ in
> country YZ)".
> 
> This list had the following advantages. A customer might argument "Nice
> that you can develop efficiently with GNUstep but what if you get
> killed tomorrow in an accident or your company goes down?" A quick
> glance on the list and I would say: "No problem, the next GNUstep
> specialized (partner) company is XXX in YYY only ZZ km away." It might
> also happen that the companies internal development capacities aren't
> sufficent for a given project. So a quick glance on the list -
> availability of developers - and the capacities are available. Another
> thing, there might be a still missing feature in GNUstep - not yet
> implemented part of the OPENSTEP/Cocoa specification - e.g. NSImage
> drawAtPoint:... or support for NSComboxes in NSTableColumns. If there
> is a real need for such a feature - project needs to be completed -
> there is often also money available to get this working, which means
> there is money available to improve GNUstep.
> 
> • Company XXX needs NSImage drawAtPoint:... working and offers YYY $
> for the job to be completed
> • Freelancer YYY (might live and work anywhere on earth) offers to do
> the job "immediately" for this amount of money
> • Feature is integrated in the GNUstep source tree and the project of
> company XXX can move on
> • All are happy
> 
> I would like to see a website - ideally powered by GNUstep -  that is
> able to bring all these pieces together. We would volunteer to do it
> but if anybody else has other suggestions...

Yes.  An excellent suggestion; I'd like to see what can be done about
putting something along these lines in to place.  Our biggest barrier
here is the fact that most commercial GNUstep applications are
probably back-end, in-house jobs, where there's little reason to
mention them outside of the circle within which they're being used and
developed.  Hopefully, those in this situation will realize the
benefit to everybody of making their work more visible (even if not
making it available).

> Just my 2 cents...

And a very valuable 2 cents they were.  Thank you :)

> Regards,
> 
>     Andreas

And likewise,

Michael Baehr