Hi Lars,
Am 18.02.2010 um 00:21 schrieb Lars Sonchocky-Helldorf:
Hi Nikolaus, hi Sergey,
my plan is as follows:
At I'll try to whip up a WebObjects based prototype (as I am
currently much more "fluid" using the tools for WebObjects and
because I never did any GSWeb before) during the next weeks which
then can easily be ported to GSWeb (at least that's what I hope for).
I also tend to use PostgreSQL as database since I consider
PostgreSQL much more sane than MySQL from my experiences.
I think we can choose any SQL database that best fits our purposes.
As long as it is SQL it should not be difficult to transfer the
existing data.
Am 17.02.2010 um 20:54 schrieb hns@computer.org:
Let's start with the data model. So, how does this work? As far as I
understood the overview one has to start a project and define the
data
model first.
Almost. Usually the very first step is to define what features the
proposed application should have, how it's supposed to work and
function, e.g. some sort of concept. I guess since we already have
an precursor here http://www.gnustep.org/softwareindex/ the
necessary effort shouldn't be to extensive (although I must admit
I don't fully understand the existing application - but then again
it's already late in the night)
Ok. Let me try to define:
"SWI is a News Feed with Archive (i.e. App-Store) for GNUstep
related Software".
Features:
* "home page" with latest news (or if you prefer other words:
announcements, publications, entries, versions)
* searchable archive of all publications (this represents the App-
Store aspect of SWI)
* items can be tagged/labeled by categories as "Developer Tool",
"Application", "Framework" etc.
* links to older (and newer) versions of the same application are
available
* each News entry contains a lot of information e.g.
- App-Name
- version number
- author
- licence
- short description
- long description (html)
- links to source code, home page, binaries etc.
-- tries to check if link works (by sending a http HEAD request)
- screenshots can be uploaded
- comments can be added by users
- ...
- not yet but should be added:
-- dependencies
-- build instructions (for automatic install)
-- (hidden) e-mail address of author to allow bidirectional
communication (e.g. if link is no longer working)
* a RSS feed of the latest news is available
* the feed can be fetched as a Property List so that it can be
analysed by an Installer tool to find new versions of installed
software and to find dependencies
* user registration
* collaborative approach: anyone can post change-requests to
existing or new entries; moderators can approve/reject them (to
protect against spamming)
* subscription and e-mail notification to individual updates
* statistics
I hope this is complete...
But right after this you usually create the data model (I use
usually EOModeler or Entity Modeler for this since those allow me
to generate the entity classes and the necessary SQL in one stop)
I think it is not very complicated as the main data structure is a
list of news entries. Additional one-to-many relationships are
needed for e.g. tags/categories, licences, dependencies, comments,
screenshots etc.
What I think we should start with the data model. Currently, we
have
these tables:
I think most parts are self-explaning when one goes through the Web
forms of the current Soctware Index.
Screenshots are not stored as BLOBs but in files on the server.
And,
we never store binaries but only links to sources and homepages.
Unfortunately the attachment was lost. Here it is inlined:
# phpMyAdmin SQL Dump
# version 2.5.6
# http://www.phpmyadmin.net
#
# Host: mysql.gnustep.org
# Erstellungszeit: 16. Februar 2010 um 13:24
# Server Version: 4.0.20
# PHP-Version: 4.3.8
#
# Datenbank: `gnustep`
#
I'll comment on that tomorrow when I am not so tired and when I
have explored the current software index a little more.
Oh yes, your mail is just 8 hours ago :-)
BR,
Nikolaus
_______________________________________________
Discuss-gnustep mailing list
Discuss-gnustep@gnu.org
http://lists.gnu.org/mailman/listinfo/discuss-gnustep