On Tue, Jul 14, 2009 at 5:28 PM, Evan Martin wrote:
> On Tue, Jul 14, 2009 at 2:24 PM, Ant Bryan wrote:
>> This is an invitation to people from Chromium and other browser/
>> download application makers to collaborate on Metalink.
>
> This sounds pretty cool, but I suspect you know more about what is
> needed to make this work than we do, and also more motivation to make
> it happen. I think if you brought up a design questions like in
> http://groups.google.com/group/metalink-discussion/msg/2c63ced761bc95e6?hl=en
> on this mailing list you'd find some helpful answers.
Thanks for replying, Evan!
>From the comment on the bug, mar...@chromium says "I wouldn't object
to a contribution so I'll leave this issue available."
We'd like help on where in the code to start contributing. :)
I've copied those design questions at the end of this mail, if anyone
has any guidance to offer, thanks!
Here's the basic flow of a metalink download:
User clicks on a URL to a metalink like
http://openoffice.mirrorbrain.org/stable/3.1.0/OOo_3.1.0_src_core.tar.bz2.metalink
metalink (XML) is downloaded.
depending on file extension (.metalink) or MIME type
(application/metalink+xml) the metalink is parsed for url element (by
libxml2?)
file from URL from url element is downloaded.
when download is completed, file is checksummed (NSS?) and compared to
checksum in hash element.
This is an important issue to us & there is a $300 bounty for the
feature if anyone's interested:
http://groups.google.com/group/metalink-discussion/web/gsoc-ideas
We're also happily accepting review/feedback on the Internet Draft:
http://tools.ietf.org/html/draft-bryan-metalink
thanks again!
--
(( Anthony Bryan ... Metalink [ http://www.metalinker.org ]
)) Easier, More Reliable, Self Healing Downloads
From: Bram Neijt
Subject: Google Chrome intergration
Hi everybody,
I've taken a evening to stare at the Google Chrome browser code and
below are some of the observations I've been doing.
I've tried to find a good place to put the code that would handle the
metalink and I haven't come up with the best possible solution yet (as
far as I can see). Has anybody else had a look at the code, or are there
thoughts floating around on this? Here's hoping somebody has better or
more ideas on this ;)
Bram
BEGIN NOTES: very sketchy, but if you saw the code you might
understand. Crosses ([X]) are not an option, [?] means possible but not
great.
Possible inserts:
[X] New browser url handler (BrowserURLHandler)
Set a new browser url handlers
Pro: will allow for new rendering scheme of XML
Con: Won't always work, because it depends on the mime, not the
extension or scheme
[X] New URLRequest
Create a new URLRequest which makes sure that any URL over HTTP is first
mime-type sniffed (by the mime_sniffer_proxy) and then checks for the
application/metalink mime-type.
Pro: This should work.
Con: What to do with a metalink with multiple files?? This would only
result in a single stream, which can't support multiple files.
[?] new DownloadItemView
Go into the DownloadItemView and add a button: when a metalink is
downloaded, allow people to push a button to start downloading the
separate files.
Pro: This would give people an option, but might leave them hanging.
Making it automatic should not be a problem
Con: This doesn't include the possibility of downloading from multiple
streams to the same file yet.
[?] onDonwloadComplete somewhere??
Maybe there is a hook which will allow us to spawn new downloads when a
metalink is downloaded completely: after download, inspect the file and
start downloads for the files next to the metalink. Not sure how to
handle the whole restart thing, but maybe some country(UTC
offset)/language checking and randomness would help.
--~--~-~--~~~---~--~~
Chromium Developers mailing list: chromium-dev@googlegroups.com
View archives, change email options, or unsubscribe:
http://groups.google.com/group/chromium-dev
-~--~~~~--~~--~--~---
