gEDA-dev: Gschem and Cairo graphics library

[Stuart Brorson]

Alas, another top post!  Sorry!

What you say is fine.  I have heard from many people over the last two
years that the install CD is brain-dead, wrong-headed, and/or just
plain stupid.  Fair enough.  But apparently enough people have used it
to install gEDA and become productive that it has served its purpose.
And that's my goal, not producing an installer which conforms to the
package-management philosophy du jour.

My challenge to those who don't like the CD is this:

Please produce a single download file which installs the whole set of
gEDA tools onto the computer of a naive user.  The tools to install
should be at least those available on my CD, if not a superset of
those tools.  The installation method can be distro-specific (i.e. one
for FC, one for SuSE, one for Debian, etc)  It can be something like a
yum server bundled with all the relevant dependencies.  Make it easy
for the naive user to get the software onto their boxen with one or a
few clicks.  And not just the gEDA/gaf stuff, but also PCB, Icarus
Verilog, ngspice, gnucap, gerbv, etc etc etc.

Don't, however, make a tool which requires an internet connection,
knowledge of Klik, or the ability to use [apt-get, yum, etc], or
require more than basic-level Linux ability.  Please keep in mind that
many users are not developers, and don't feel comfortable with (or
even know about) these package/dependency management tools.

(I gotta say that I field a lot of support requests since we run the
Free Dog meetings in Boston, and since I produced the install CD.  The
level of questions/problems we have with users has convinced me that
folks who say "just use [apt-get, yum, etc]" simply don't get it!
Knowledge of these tools is not as widespread as you might think!  One
thing Windoze does well (as opposed to Linux) is that installation of
software using wizards is trivial.  (Yes, the Windoze package may end
of being strangely misconfigured at some later date, but that's
invisible to the users, so they are happy with their install
experience!)  Unfortunately, prospective gEDA users are very
frequently flummoxed by how to install gEDA, and it is important for
us developers to be sensitive to this problem.)

Getting back to my challenge: If somebody would step up and produce
such a beast, I am 100% sure that his installer would end up in a
prominent place on the gEDA downloads page.   It would be a great
thing for the gEDA project!  And I'd be happy to see my CD Installer
become just a happy memory if we got a complete set of more 
spiffy installers.   However, unless somebody actually produces such a
beast, all opinions on this issue are "full of sound and fury,
signifying nothing."

Stuart


On Sun, 30 Jul 2006, Peter TB Brett wrote:

> On Sunday 30 July 2006 13:53, Stuart Brorson wrote:
>
>> Developers say "just use [yum, apt-get, etc]" to get the dependencies,
>> but our target audience is not only other software developers who know
>> these tools, but also ordinary users -- students, hobbiests,
>> technicians, working engineers, and so on.  They are very often almost
>> completely clueless w.r.t. administering a unix system, and are
>> quickly put off by our software when they can't build it (and don't
>> know how to handle the dependency issue).  From reading the
>> complaints on this list as well as on the web, I've concluded that we
>> lose a good number of potential users before they ever even draw a
>> schematic since they can't install gEDA.  And the biggest reason they
>> can't install is the dependency problem.
>
> I still maintain that if someone can work out how to use a graphical package
> manager to install regular software, it's not overly optimistic to assume
> that the can use the same graphical package manager to install the
> relevant -devel, -dev or [whatever else they happen to be called] packages.
>
> Cairo would be a very great aid in simplifying the addition of useful export
> filters to libgeda, as well as in making the renderer code more stable and
> easier to maintain.  The current gschem rendering code has problems:
> corruption of the active view occurs far too often, and it can get very slow
> for complex schematics.  Using Cairo's (transparent) hardware acceleration
> would be fantastic.
>
> I personally see no objection to adding a library dependency, provided that it
> addresses an actual need (in this case it does), and that distributions can
> be reasonably expected to provide it in the core package set as opposed to in
> an 'extras' repository (it's in core on FC5 and Ubuntu, at least. I haven't
> checked other distros).
>
> The fact it might make you CD not work quite as well doesn't really bother me
> hugely, I'm afraid.  Personally, I think it's a flawed concept.  The target
> audience for the CD (x86 Linux users who don't know how to build software by
> themselves) would be much better served by a klik package
> <http://klik.atekon.de/>.  Firstly, they're likely to be using
> a 'newbie-friendly' distro such as (K)Ubuntu, Freespire/Linspire, SUSE,
> Knoppix or FC5, most of which are well-supported by klik.  Secondly, it
> doesn't involve compiling software.  Thirdly, it doesn't involve installing
> system packages that might well be better installed from the package manager.
> Finally, klik packages are installed as non-root, and are easily uninstalled.
> I think that your extremely impressive effort might well been better spent
> working on that, rather than the CD.
>
> I also think that the wiki installation pages are wrong.  At the moment, the
> easiest way to install the gEDA suite is to use the RPMs/DEBs/Fink packages
> that people have so kindly built, and get the package manager to sort out all
> the dependencies for you.
>
> Of course, the best thing would be to have geda in distros' 'extras'
> repositories, but as you rightly point out it's quite hard to find
> maintainers for that kind of package.
>
> <flame>Having said all that, telling developers who've worked hard on
> something of genuine use to go take their patches and shove it because it'll
> make your life slightly more difficult is really poor form.</flame>
>
> Regards,
>
> Peter
>
> P.S. The wiki links in README-CD are out of date.
>
> -- 
> Fisher Society publicity officer            http://tinyurl.com/o39w2
> CUSBC novices, match and league secretary   http://tinyurl.com/mwrc9
> Quake II build tools maintainer             http://tinyurl.com/fkldd
>
> v2sw6YShw7$ln5pr6ck3ma8u6/8Lw3+2m0l7Ci6e4+8t4Eb8Aen5+6g6Pa2Xs5MSr5p4
>  hackerkey.com
>