The Browser Problem...

[Anthony Bryan]

we've accomplished a lot, but without support in browsers, even the
ones w/ less users, we haven't really reached some of the people who
need metalink's features the most. they will never install plugins or
external download managers.

so far, we have TheWorld browser. this is a great start, but we need more.

Opera, Safari, Google Chrome, & Firefox would all be great to have on board
what's holding them back? not enough sites? why aren't more sites
using it? almost no browser support :)

what can we do to speed things up?

we have bounties for the open source browsers, Chrome & Firefox, &
they're both waiting for someone to submit patches.

Shawn Wilsher, in charge of Firefox's download manager code, is at
least thinking about it:
"Here's what I'm thinking the right approach to do this properly goes
something like the following. Note that for coming up with this, I'm
basically thinking how we could implement metalink downloads in an addon. " [1]

http://discussions.apple.com/thread.jspa?messageID=4944688&#4944688

http://my.opera.com/community/forums/topic.dml?id=152871

http://code.google.com/p/chromium/issues/detail?id=1751

https://bugzilla.mozilla.org/show_bug.cgi?id=331979

[1] http://groups.google.com/group/mozilla.dev.platform/browse_thread/thread/9247f53da4527422#
--
(( Anthony Bryan ... Metalink [ http://www.metalinker.org ]
)) Easier, More Reliable, Self Healing Downloads

[Bram Neijt]

Well, I personally think that more support from sites would help allot.
One problem is, however, trying to get the sites to create them.

It is almost trivial to create a library, function or patch that creates
them, but getting them to be used is the hard part.

If needed, I could write a bouncer like system completely in javascript
on the client side:
- AJAX the metalink
- Check time zone of local user against country timezones
- Randomly select from 5 nearest servers

All of that would probably work great and you could also just show a
sorted list of nearest servers: all sorting, thinking and matching can
be done on the client side.

The problem however, is to find somebody who would be willing to use my
code or with a specific place to put it. I'd love to see some admins
coming forth and telling me: I'm using Django/PHP/Python/Javascript and
would like to incorporate metalinks into my download page, can you guys
help me with some code to do that?

Now implementing it on the browser side, without a separate client,
seems much harder. Although an algorithm to select a single link from a
metalink is easily created, it won't solve the problem of possible
failures or repeats. As far as I have been able to gather from Firefox
and Chrome is that there isn't anything that does this: if a download
fails it just fails and shows the user it failed. Adding a metalink
payload to the download seems difficult without some real structural
changes to the download managers of the browsers :S.

So with what little time I have left, I'll help pushing on the server
side as much as possible.

Bram

On Wed, 2008-12-10 at 02:37 -0500, Anthony Bryan wrote:
> we've accomplished a lot, but without support in browsers, even the
> ones w/ less users, we haven't really reached some of the people who
> need metalink's features the most. they will never install plugins or
> external download managers.
>
> so far, we have TheWorld browser. this is a great start, but we need more.
>
> Opera, Safari, Google Chrome, & Firefox would all be great to have on board
> what's holding them back? not enough sites? why aren't more sites
> using it? almost no browser support :)
>
> what can we do to speed things up?
>
> we have bounties for the open source browsers, Chrome & Firefox, &
> they're both waiting for someone to submit patches.
>
> Shawn Wilsher, in charge of Firefox's download manager code, is at
> least thinking about it:
> "Here's what I'm thinking the right approach to do this properly goes
> something like the following. Note that for coming up with this, I'm
> basically thinking how we could implement metalink downloads in an addon. " [1]
>

> http://discussions.apple.com/thread.jspa?messageID=4944688?

[Anthony Bryan]

thanks Bram! I don't know where efforts should best be focused.

as you (kindly) put it, few people are asking for help using metalinks
on their sites. sounds like an education/PR/marketing problem.

I mean you patched Bouncer, what almost 2 yrs ago?! but they didn't
use it. if the code is there, but unused, that's a shame. we are only
trying to help them.

one thing we've talked about is spreading the word at conferences. you
gave a lightning talk at FOSDEM a few yrs ago. do you think it's worth
it for someone (not necessarily you) to try to go back? do you think
there would be a better reception now?

http://www.fosdem.org/2009/call_for_lightningtalks 2008-12-26:
Deadline for lightning talk requests

back on the browser front, I heard from Paweł Hajdan jr who is the
first person from the Google Chrome community with commit access.
Paweł said that he would help address issues during review or
committing process if we contribute patches adding metalink support.
he isn't able to do patches right now. also, the extension mechanism
is in progress. this could be nice, but we really want basic metalink
support in Chrome natively to reach the most people.

--

[Anthony Bryan]

an invitation to Chromium & Firefox, & at least a reply from Chromium!

http://groups.google.com/group/chromium-dev/t/2ce5e33a1689c379

http://groups.google.com/group/mozilla.dev.platform/browse_thread/thread/7bc5ab6629b0184d#

2008/12/16 Anthony Bryan <anthon...@gmail.com>:

[Anthony Bryan]

some favorable comments from Mozilla/Firefox

http://groups.google.com/group/mozilla.community.web-standards/topics

2009/7/14 Anthony Bryan <anthon...@gmail.com>:

[Anthony Bryan]

is there no one with C++ experience who can add basic metalink support
to a browser?

I think Google Chrome is the best target. they're new & developing
their browser rapidly.
if we can supply the code, they're review it & add it.

http://code.google.com/p/chromium/issues/detail?id=1751

[Bram Neijt]

Some time back I looked at the download systems, like Chrome, but the
biggest problem was that the main download engine does not support
multiple sources. I couldn't find a way to add multiple download
sources, but it may be a good time to try it again.

I'll see if I can find the time to check it out and start a new thread
on implementing this in either Firefox or Chrome.

Bram

PS I did a quick survey on Chrome earlier:
http://groups.google.com/group/metalink-discussion/browse_thread/thread/24de58f2c4b7785c/2c63ced761bc95e6
but at that time I could not find a good way to hook into the download
system. Hopefully there is more documentation now or maybe somebody can
find a good way in.

[Anthony Bryan]

thanks for investigating, Bram.

when I say basic metalink support, I mean: get one URL from the XML,
download from one source, and verify the file's checksum.

most apps that download from multiple source support metalink at this
point. to get our foot in the door with other apps, I think we need to
be very simple. this means less benefits to users, but we proceed with
small steps...then, the main program authors (or users, once they see
basic functionality) might find out about metalinks & find em
interesting enough to add more features, like failover to other
mirrors. many apps will never download from multiple sources.
hopefully they'll add chunk checksumming support later.

[Hampus Wessman]

Last time I looked at Chrome, it could only be compiled with Microsoft Visual Studio (not the free version). That's a huge obstacle for all open source developers who mainly runs linux and don't want to spend money on expensive Microsoft compilers.

I got the impression that it wasn't going to change any time soon either. I like the browser, but I just can't understand why they make it open source and then don't bother to make sure you can compile it with a free (as in beer at least) compiler. That excludes many potential developers. Does anyone know if it's still like this?

Hampus

[Anthony Bryan]

sounds like they are making progress with this, if Visual C++ 2008
Express Edition is the free version

http://cotsog.wordpress.com/2009/11/08/how-to-compile-google-chrome-with-visual-c-2008-express-edition/

[Tatsuhiro]

On 11月16日, 午後11:38, Hampus Wessman <hampus.wess...@gmail.com> wrote:
> I got the impression that it wasn't going to change any time soon
> either. I like the browser, but I just can't understand why they make it
> open source and then don't bother to make sure you can compile it with a
> free (as in beer at least) compiler. That excludes many potential
> developers. Does anyone know if it's still like this?
>

I compiled chromium few days ago on Debian/sid amd64. There is a
documentation to how to build in linux.
http://code.google.com/p/chromium/wiki/LinuxBuildInstructions
During build, I didn't come up any compile error :)

[Hampus Wessman]

Cool! Nice to hear that things are getting better :)

[Hampus Wessman]

Yep, that's the free version. Nice!

[Nicolas Alvarez]

Hampus Wessman wrote:
> Last time I looked at Chrome, it could only be compiled with Microsoft
> Visual Studio (not the free version). That's a huge obstacle for all
> open source developers who mainly runs linux and don't want to spend
> money on expensive Microsoft compilers.

People who mainly run Linux can compile it for Linux using gcc.

"Only supporting Visual Studio" is true for Windows. (I don't think the
Windows version compiles under MinGW)

[Hampus Wessman]

Last time I looked at Chrome you couldn't compile it for Linux at all and it needed some non-free Microsoft library on Windows (ATL, I think). That's the main reason I haven't even considered developing anything for Chrome. Really nice that the Linux port is making progress! Being able to help out with the development without even having to leave Linux is perfect.