Bad Content-Length and page never finishes loading

VERIFIED FIXED in M15

Status

()

defect
P3
normal
VERIFIED FIXED
20 years ago
20 years ago

People

(Reporter: colin, Assigned: rpotts)

Tracking

Trunk
Points:
---

Firefox Tracking Flags

(Not tracked)

Details

(Whiteboard: [PDT-])

M11 code on both Linux and OpenVMS platforms. Probably occurs on all platforms.
Still happening in my M12 build (code pulled around 11/24).

If the server lies about the Content-Length (says there's more content than
there really is), then Mozilla never finishes loading the page. While you could
argue that this is an invalid scenario, it does happen in the real world, and
other browsers (such as Navigator) handle it just fine.

You can see this problem on:
http://www.openvms.digital.com/openvms/products/ips/mozilla_relnotes.html

The server says the content length is 23172, but it only sends 22147 bytes of
content down the wire.

I have reported this problem to Compaq and I'm sure they will be fixing the bad
page soon. I have asked that they rename (rather than delete) this bogus page
when they fix it, and I will post the new URL of the bad page when that happens.
The bad page is now:
http://www.openvms.digital.com/openvms/products/ips/mozilla_relnotes_broken.html

The "good" page hasn't been fixed yet, either.
The real page is now fixed. You'll need to access the "broken" URL to
see this problem.
Assignee: gagan → rpotts
Summary: Bad Content-Length and page never finishes loading → [dogfood] Bad Content-Length and page never finishes loading
Target Milestone: M12
Rick, can you take a look?
Whiteboard: [PDT-]
Putting on PDT- radar.  We would need to see a major site that shows this
problem.
Moving milestones...
Bulk move of all Necko (to be deleted component) bugs to new Networking

component.
Assignee: rpotts → gagan
We're going to have to honor content-length to implement keep-alive (rather
than relying on the server closing the stream). Giving this to Gagan.
Target Milestone: M13 → M14
Keywords: beta1
Putting dogfood in the keyword field.
Keywords: dogfood
this should timeout from the sockettransport... right? off to rick.
Assignee: gagan → rpotts
Summary: [dogfood] Bad Content-Length and page never finishes loading → Bad Content-Length and page never finishes loading
In my M14 build, this problem doesn't occur any more. A page which is shorter 
than content-length DOES finish loading. Someone fixed it!!
I don't know what it could have been that caused it to start working. Did the 
socket transport or http change?

Maybe we should consider this closed.
Target Milestone: M14 → M15
It'll time out now.
Status: NEW → RESOLVED
Closed: 20 years ago
Resolution: --- → FIXED
verified:
NT 2000042009
Linux 2000042609
Mac 2000041810
Status: RESOLVED → VERIFIED
You need to log in before you can comment on or make changes to this bug.