On Sunday 11 April 2010 02:04:17 pm, Ulrich Dangel wrote:
IMHO it should really be more like a wizard, e.g. select 32/64bit, the flavor (small/medium/normal) and then probably (?) the version, stable, rc, daily images.
And if you want you can also allow to choose the base distribution for the daily images (stable/testing/unstable)
Ulrich, I hope you don't mind me hijacking your reply, I just subscribed to the list recently and don't have the original message to reply to.
I haven't finished my mock-up yet but it is complete enough to give people a general idea. See:
http://caligula.rhombic.net/~xjjk/grml/
I'm making this with the use cases I have in mine as a grml user. I really don't think a Ubuntu-style wizard is useful or appropriate (honestly, I've never used that page--I've always gone directly to cdimages.ubuntu.com for getting Ubuntu CD images), especially given grml's technically oriented audience. Rather than a wizard, my mockup is intended to be more a "filter."
My thoughts on the ideal download page (most of this is in the mockup):
* Filename should be visible. I want to know exactly what I'm downloading. * Basic file metadata should be visible, with the minimum being an image's SHA1. With the current site, you have to read a lot of text and click 3 or 4 things to reach it. I included file size and MD5 in the mockup, I'm not sure whether these are necessary. * Architecture, Flavor, and Version filters should be automatically selected to the most popular download. * All links/information about a CD image should be grouped together, including metadata, release notes, and direct download/BitTorrent/rsync links. Download links should point directly to files (including rsync--since I don't think the new DNS name handles rsync I'm not sure the best way to do this). If I want to download an image file to say, my headless download server, it should be no more than 2 clicks (copy link from download page, paste into terminal). Same for BitTorrent and rsync.
On technology (at the time of this e-mail this is not evident in the mockup):
* At release time, a script should generate/collect all the metadata about each CD image. This should get output as a new HTML file. This script can be either be run automatically, via some kind of event hook, or manually by the maintainer. Said HTML page is static (w.r.t. the server). * Said HTML file should contain information about all releases (or perhaps only 2 or 3 releases back). Javascript (jQuery) will be used to filter through the releases people want (I'm sorry, it's 2010: if you're still avoiding Javascript, then just deal with the long list).
I'm going to continue working on this for the next few days, in particular adding Javascript to make the filtering work (unless the response is that everyone hates it). Please let me know what you think!
Regards, Samat