17067 - Webcam32 JavaScript "Image Pull" does not work

Status: VERIFIED DUPLICATE of bug 29643

(Core :: DOM: Core & HTML, defect, P3)

Description

[WD]

Overview Description:
With the Webcam software package, Kolban Webcam32, there is a new technique
called Image Pull.     Previous versions used the JPEG push technique, which
worked fine for Netscape, but Internet Explorer would not work with it.   Well,
this Image Pull technique uses JavaScript to work around this, creating one
solution that would work for both Netscape and Internet Explorer.

Unfortunately, it does not seem to work at all with Mozilla.
A web page describing Image Pull is available at:
http://surveyorcorp.com/webcam32help/imagePull.htm


Steps to reproduce:
Visit:    http://a2051.cafaro.ysu.edu/index1.html

Actual results:
No webcam picture is displayed

Expected results:
"streaming" webcam picture is displayed, as with Netscape or Internet Explorer

Build Date and Platform:
Mozilla 1999102110
Windows 98, 95

Additional Information:
I believe the above sample page should suffice, but if you want to do some
heavy-duty testing, you can download and install Webcam32 on your computer, and
visit the ImagePull web page above to test ImagePull from your own computer.
But still, It'd be MUCH simpler to visit http://a2051.cafaro.ysu.edu/index1.html
, since it's already set up and running.

I'm not sure if the problem here is with the implementation of JavaScript, or if
this is just a feature which hasn't been added to Mozilla yet or what...

-WD

Comment 1

[Mike McCabe]

WD -

Interesting bug!

Could be something with our network stuff, or it could be some other aspect of
your page.  I'm reassigning the bug to the networking-core component, in the
hopes that someone there will recognize it.

Really, though - your best chance for getting this resolved is to pinpoint the
exact problem, or to narrow down your testcase to the smallest possible example.
Looks like you understand the something about our network streaming; can you
provide an example that shows that we're doing something wrong?  Or, if it's
cross-browser code, maybe there's something in your browser-detection code that
makes it work, that's not working with mozilla.  We get lots of bugs :) so any
help you can offer makes it much more likely we can get this fixed.

Comment 2

[WD]

Attachment: Source code for sample webcam page using "Image Pull" (or push)

Comment 3

[WD]

Ok, here's some more information about this issue:

As it turns out, not only does Mozilla not support this "Image Pull" feature,
but it does not support JPEG server push, either.    JPEG push has only been
supported by Netscape in the past, but I would think that this would be
something to implement in Mozilla, also.

For a sample page with JPEG push, check out:
http://www.ecn.purdue.edu/ECN/Webcam/right

With Netscape, this page works fine and displays a streaming image.  With
Mozilla, it appears to be constantly downloading information, but the picture is
never shown.    Internet Explorer just shows a broken image on this page, and
just stops there.

As for browser detection, take a look at the ZIP file above.  This is the source
code for a basic Image Pull page.    It does check which browser is in use by:
browserType = navigator.appName;

If it's "netscape", it uses the standard server JPEG push.
If it's anything else, it uses the new "Image Pull" technique designed by the
creator of Kolban Webcam32, I believe.    This technique uses JavaScript to
simulate JPEG push in an environment where it is not supported:  It loads an
initial image, and as soon as that's done, the client PC requests a new image
and displays that (and loops..)

I'm no Javascript guru, but take a look at the ZIP file to see what exactly is
going on...   (it doesn't seem too complex to me)

Following this message, I will post the source code for the page
http://a2051.cafaro.ysu.edu/main.html            (which uses Image Pull)
It is virtually the same as the ImagePull sample ZIP file above, except with the
appropriate server name inserted.     It should give you a nice test case to
work with.

Comment 4

[WD]

Attachment: Source for Main frame of live page using ImagePull script

Comment 5

[cbegle]

updating qa contact to tever.

Comment 6

[Gagan]

ImagePull is nothing more than a request for a document image thru Image().src
(The same as what we do for preloading images for JS animation) Since
individual requests work fine, I suspect something flaky in the JS part. Am
handing this back to mccabe.

Comment 7

[Mike McCabe]

I hate to do this, but -

I have no idea how images work.  I strongly doubt that it's a problem with the
JavaScript language; I'd need to see a simpler testcase showing that it was.

Lacking a better place to put it (as it's been tossed back at me) I'm going to
reassign this to the DOM component, in the hopes that it'll get some external
evaluation through Eric Krock's bugathon.

CC'ing gagan, as he seems to be involved with both networking and images, and
may be able to cast more light.

Comment 8

[Gagan]

any simple javascript involving this line should theoretically have the same
problem (haven't tried it as yet) Assuming there is an image on a document-

    document.images[0].src = "http://..." // some img location.

This should trigger either a request to the imagelib or necko for the new
location. (I think its the former) but currently necko doesn't see this request.

Comment 9

[vidur (gone)]

In an attempt to get my bug list in order again, marking all the bugs I have
currently as ASSIGNED.

Comment 10

[vidur (gone)]

The reason the page isn't working is because it tries to do server push if the 
appName is "Netscape" (as it is for Mozilla). It turns out that the client pull 
stuff won't work correctly either since image onLoad handlers aren't implemented 
- see bug 22710. 

Passing this back to gagan for the server push part. If we aren't planning to 
implement this, just DUP this bug with 22710.

Comment 11

[Laura Yecies]

pdt - we don't think server push is beta1

Comment 12

[Gagan]

for mozilla builds this should failover to the non-netscape version of the
script which relies on client-pull and would rely on bug 22710. 

for the commercial builds we should do the server push and this should work ok.I
will verify as soon as I can get a commercial up and running. 

Comment 13

[Judson Valeski]

yea, the JS should work, but the failover to the x-mixed-replace should too
(this is bug 29643)

Comment 14

[ekrock's old account (dead)]

wdormann, good catch! Since this bug now seems to be tracking the "Image Pull" 
problem, could you please open a separate bug to independently track the fact 
that JPEG server push is not working? That way, we'll make sure that both 
problems get tracked through separately to resolution. (We try to follow the 
rule "one issue = one bug" in Bugzilla, so when we discover through analysis 
that one report contains two issues, as in this case, we try to split off one 
issue into a separate report.) I'd do that myself except that I don't understand 
very well which info & URLs apply to which problem and you obviously understand 
it *extremely* well. Thanks very much for your help in bringing these issues 
to our attention and your great work on simplification!!

Comment 15

[WD]

Thanks. . .

And actually, I did file a separate bug for JPEG push.   It was bug 21976
It has been marked a duplicate of bug 6074 , though.   That's the current bug
describing JPEG push.   (Though it doesn't have "JPEG Push" in its summary)

Comment 16

[WD]

Actually, that was supposed to be bug 21796 that was marked as a duplicate.

I have posted some more information about JPEG push under the currently open bug
6074

Comment 17

[WD]

Here is a page that has the Webcam32 "Image Pull" script, but with the Netscape 
browser detection removed.    i.e., it uses Image Pull, rather than JPEG Push 
for *all* browsers, (netscape included)

http://dormcam.webhop.net/index2.html

On NN and IE, the video pumps out at about 1fps, but with Mozilla it only shows 
the original image.
(I run that site, so if anybody would like me to modify the javascript code for 
testing or anything, just let me know)

Comment 18

[Warren Harris]

Moving to M17 which is now considered part of beta2.

Comment 19

[Judson Valeski]

*** This bug has been marked as a duplicate of 29643 ***

Comment 20

[Tom Everingham]

verified Dup of 29643 

Comment 21

[Pulsebot]

Pushed by ffxbld-merge:
https://hg.mozilla.org/releases/mozilla-beta/rev/f4b304e0d6dc
[fenix] For FNXV2-17067: always show home in background behind search dialog (https://github.com/mozilla-mobile/fenix/pull/20573)