226230 - [FIX]Ha'aretz home page downloads ads as Excel files!

Status: RESOLVED FIXED

(Core :: Networking, defect, P1)

Description

[david toub]

User-Agent:       Mozilla/5.0 (Macintosh; U; PPC Mac OS X; en-us) AppleWebKit/103u (KHTML, like Gecko) Safari/100
Build Identifier: 

New with the 10/19 build, if I navigate to http://www.haaretzdaily.com/, a bunch of excel files 
suddenly download to my desktop, which seem to be ads. This behavior was seen in early betas of 
Safari, and has never been seen before with Camino. 

Reproducible: Always

Steps to Reproduce:
1.Go to http://www.haaretzdaily.com/
2.
3.
Actual Results:  
Several files download to my computer, which are interpreted as Excel files, representing ads from 
that page. If I leave the browser open for several minutes to the Haaretz home page, the same 
thing happens again.

Expected Results:  
Download nothing, just load the page and that's it

Comment 1

[Mike Pinkerton (not reading bugmail)]

yuck. dagley, what did you do to the auto d/l code that would cause this?

Comment 2

[Steve Dagley]

David, When you say "New with the 10/19 build" do you mean the 10/18 build
didn't exhibit this problem?

I don't see it in the Camino 0.7 build so it could be something that didn't get
moved from the Camino branch to the trunk.  Does the download window show an
entry for each item downloaded or do the files get downloaded with no indication
from the UI?  I'm guessing that the files downloaded are images have an
incorrect mime type making us think they've got to be downloaded rather than
being displayed on the page.

Comment 3

[Mike Pinkerton (not reading bugmail)]

they're going through the d/l window

Comment 4

[david toub]

Also happens with the latest Firebird nightly (it did not do this with the 11/13 build), so I strongly 
suspect it is something with the trunk that happened either on 11/18 or 11/19 as this behavior 
was not in Camino's 11/14 build. I had not tested 11/18.

Comment 5

[Boris Zbarsky [:bzbarsky]]

There've been a few file-handling changes recently, but they should not have had
this sort of effect...

Comment 6

[Christian :Biesinger (don't email me, ping me on IRC)]

so, er, has this started in the 10/19 build as stated in comment 0 or in 11/19
as stated in comment 4?

that said, I can confirm that mozilla brings up the helper app dialog. it shows
a content type of octet-stream.
And apparently rightly so:
0[244608]: Found extension '' (filename is
'site=haaretz&adsize=120x60&hposition=30&hlayer1=hp&hlang=ENG', handling
attachment: false)
0[244608]: HelperAppService::DoContent: mime 'application/octet-stream',
extension ''

Comment 7

[Christian :Biesinger (don't email me, ping me on IRC)]

(my testing was done with nightly 2003111908 on win2k)

Comment 8

[Boris Zbarsky [:bzbarsky]]

To further narrow the range, my CVS build from "Sun Nov 16 16:03:17 CST 2003"
brings up the helper app dialog, while a 2003-11-10-05 nightly does not.

I'm going to see whether I can locate some builds covering that range.  Damn
this "no nightlies" crap with ftp.mozilla.org.  :(

Comment 9

[Christian :Biesinger (don't email me, ping me on IRC)]

bz: maybe try backing out the bug 78919 patch... that's the first patch that
comes to mind in this area that was checked in recently...

the other uriloader changes were win-only except bug 223990 which I find very
unlikely.

oh wait
when was your checkin to detect low characters (0-32 or so) as octet-stream, was
that in this timerange/is it in your tree?

Comment 10

[Christian :Biesinger (don't email me, ping me on IRC)]

(the checkin for that detection I mentioned was 11/18/2003 07:10 PST, bug 126782)

Comment 11

[Boris Zbarsky [:bzbarsky]]

Yes, i's in my tree...  That could very well be it.  I'll test.

Comment 12

[Boris Zbarsky [:bzbarsky]]

Doesn't seem to be it.... I'm converting some tinderbox builds into tarballs so
I can just run a regression lookup...

Comment 13

[Christian :Biesinger (don't email me, ping me on IRC)]

>Doesn't seem to be it....

that is interesting. my testing has just shown that that _is_ it.

Comment 14

[Boris Zbarsky [:bzbarsky]]

Oh, doh.  I'm incapable of backing out my own patches, apparently... ;)

I'll take a look at what sort of garbage they're sending...

Comment 15

[Boris Zbarsky [:bzbarsky]]

This is so mine.

Comment 16

[Boris Zbarsky [:bzbarsky]]

This should really not end up in beta... I made a dumb mistake.  :(

Comment 17

[Boris Zbarsky [:bzbarsky]]

Attachment: Patch. I need to look at that page in K&R more. :(

Mental note.  (32 & 31 != 32) evaluates to 0, since != has higher precedence
than &.

Comment 18

[Boris Zbarsky [:bzbarsky]]

Comment on attachment 135957 [details] [diff] [review]
Patch.  I need to look at that page in K&R more.  :(

Could I have some review love here?

Comment 19

[Boris Zbarsky [:bzbarsky]]

Oh, the "garbage" is a space char, of course...

Comment 20

[david toub]

Oops, I meant 11/19 rather than 10/19. Forgot what month we're in...

Comment 21

[Darin Fisher]

Comment on attachment 135957 [details] [diff] [review]
Patch.  I need to look at that page in K&R more.  :(

whoops... i should have caught that :-/

sr=darin

Comment 22

[Boris Zbarsky [:bzbarsky]]

Comment on attachment 135957 [details] [diff] [review]
Patch.  I need to look at that page in K&R more.  :(

Could this please be approved for 1.6b?  This fixes a silly C coding error...
:(

Comment 23

[Brendan Eich [:brendan]]

Comment on attachment 135957 [details] [diff] [review]
Patch.  I need to look at that page in K&R more.  :(

gcc should warn -- does it?

/be

Comment 24

[Boris Zbarsky [:bzbarsky]]

It does not.

Checked in for 1.6b.