Re: Intent to create a much nice WWW package viewer
--On Sat, Jul 4, 1998 7:53 pm -0400 "Kevin Atkinson" <kevina@clark.net>
wrote:
> Hi there,
>
> Remember Me.
>
I do.
> Well this time I actually plan on doing something. I plan on making a
> very very nice WWW interface for retrieving random packages (something I
> do all the time).
Cool idea. Some comments, though.
First, generally. Your general game-plan seems to be a clever package
downloader, with a nice interface. What you are doing seems to me to be
something rather like a replacement for dselect - and your description
suggests to me that you are someone who uses apt, but not dselect, and
wishes for some of the dselect functionality. We already have a project to
do this - and it's called apt.
The apt you see at the moment is simply the dependency resolution,
downloading, and install-time package ordering system. The entirety of the
apt project will include a smart new user interface which is capable of at
least some of what you describe.
>
> Here is my game plan
>
> 1) Import all the package file in (bo, hamm, slink, project, Incoming)
> into a SQL database. I am using an SQL database to greatly eaze things
> in the long one and to make it very eazy to create powerful searching
> faculties as well as being able to view package information in any
> number of formats.
A few specifics.
1) Use of Incoming should be very optional, with the default being off.
2) There is already a project in the works to move dpkg's internal databases
(currently text) to a binary format. Or, at least, Ian Jackson has heavily
hinted that such a change is in the offing on several occasions.
3) You mean facilities, not faculties. Faculties are quite different ;-)
>
> 2) Create the web interface for just looking at the packages and just
> downloading that packages
>
> 3) Work on being able to select more than one package at a time and work
> on a way to display all dependices (not just the ones directly related).
>
> 4) Here comes the big one. Work on a way of finding out the setup of
> your debian configuration and work out any conflicts/dependencies.
>
[ingenious perl-server method deleted]
I imagine that could me made to work. It's very clever - but what you have
adequately demonstrated is that your project is not really ideally suited
for the WWW. In general, one should bear in mind that HTTP is primarily a
one-way information access protocol.
I am not quite sure what advantages your project would have over the GUI
version of apt, which I suspect is only a few months away anyway (Jason?)
>
> Unless I get an account on one of those machines I plan on doing the
> work at http://sunsite.unc.edu/kevina/debian/ using MySQL as I already
> have an account on sunsite and they have a mirror of the debian
> distribution (although hamm contains a symbolic link to itself)
As an aside, all developers have accounts on the debian machines, and
accounts are very rarely given to non-developers. So if you're serious
about projects of this scope, I suggest you sign up as a developer.
>
> Cgi programming and organizing information is one of my strong points.
> Go to
> http://sunsite.unc.edu/kevina/ to look at some of the work I have done.
>
> Thanks again,
>
> And please be nice.
I'm always nice ;-)
Jules
/----------------+-------------------------------+---------------------\
| Jelibean aka | jules@jellybean.co.uk | 6 Evelyn Rd |
| Jules aka | jules@debian.org | Richmond, Surrey |
| Julian Bean | jmlb2@hermes.cam.ac.uk | TW9 2TF *UK* |
+----------------+-------------------------------+---------------------+
| Debian GNU/Linux - "Microsoft *does* have a year 2000 problem - |
| and we're it!" (paraphrased from IRC) |
\----------------------------------------------------------------------/
--
To UNSUBSCRIBE, email to debian-devel-request@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org
Reply to: