On Mon, 2011-09-19 at 22:04 -0400, Hongqing Yang wrote:
----- Original Message -----
> From: "Kamil Paral" <kparal(a)redhat.com>
> To: "AutoQA development" <autoqa-devel(a)lists.fedorahosted.org>
> Sent: Monday, September 19, 2011 9:58:11 PM
> Subject: Re: iso watcher, post-iso-build event, mediakit_sanity test patch for
review
>
> > Install TCs/RCs should always be in
> >
http://dl.fedoraproject.org/pub/alt/stage/*/Fedora/, and lives in
> >
http://dl.fedoraproject.org/pub/alt/stage/*/Live/. Would it be
> > enough
> > to just
> > ignore everything else?
>
> It is doable, but submerging into individual directories is not
> optimal, because it is accessible only through HTTP. It's not really
> a one-liner. And more code -> more failures. But definitely a way to
> go in the worst case.
>
The lftp also has a --include function, I have tried it, it does not work, I will
continue investigating it.
And James has written a script to download the iso before, I also updated it as the two
stage download script,
That uses the HTML parser libxml2 and find all the uris ends with 'iso', then
download them, That is a possible
solution, but the code is complicated.
We can also mirror all the files at dl.fp.o, but we have to write script to clarify the
categories, such as
DVD, netinst, Live, deltaiso, etc.
At the time, and still today, we don't have a standard format for ISO
drops. We know the server, and the root directory where things will
land ... but the directory structure and names change each
drop/milestone/release. We can work to define some consistency there
with Fedora release engineering. However, the quick'n'dirty
proof-of-concept approach I took was to talk the directory tree (with
safe guards) to locate applicable content for download.
As Hongqing notes, it does add some complexity over a simple `curl`
command. But still seems sustainable and the quickest option IMO.
Thanks,
James