[Edbrowse-dev] Continuing Amazon issue

[Edbrowse-dev] Continuing Amazon issue

[Chuck Hallenbeck]

Hi everyone,

My Amazon sign-in problem is shared by Larry Baggett, so I'm
copying him on this.  I am including the final dozen or so lines
following submitting my sign-in request with db3 selected.
A script of the failed sign-in effort shows more, it is at:
www.panix.com/~chuxroom/signing-in.txt


Here are the last dozen or so debugging lines:

content application/x-javascript
mod date Mon, 16 Oct 2017 16:18:16 GMT
http code 200
from cache
execute fwcim._CB513135890_.js at 1
execution complete
execute signin at 571
execution complete
execute signin at 592
execution complete
execute onload
execution complete
444
TypeError: undefined not callable (property 'querySelectorAll' of [object Object])
execution complete
TypeError: undefined not callable (property 'debug' of [object Object])
execution complete



Befuddlement also complete! Anybody have any idea? Karl says his Amazon 
sign-in works, but Larry's doesn't, and mine doesn't.


Chuck


-- 
Here In Northeast Ohio also, The Moon is Waning Gibbous (61% of Full)
When your only tool is a hammer, everything looks like a nail.
Sent from Ashley's iPhone.

Re: [Edbrowse-dev] Continuing Amazon issue

[Kevin Carhart]

Hi Chuck, Larry & list,

I hope we can do something on this, but it's more difficult than usual.

Dominique said:
> I'm not sure I'll be of much help but I'll try to look at it, even if
> the code is obfuscated it should still run in duktape...

I think the current situation is a runtime error or 
something unhandled from within the obfuscated file.  That's my reason for 
thinking that we have to dig in.  It's true that we toss endless amounts 
of js files from pages into duktape, which work beautifully, and we don't 
want or need to care about that JS code.


> TypeError: undefined not callable (property 'querySelectorAll' of [object 
> Object])
> execution complete
> TypeError: undefined not callable (property 'debug' of [object Object])
> execution complete

Another tricky thing is that these don't have line numbers attached, so 
I'm assuming they are from event handlers, which are no longer in the form 
of readable JS.  We've mentioned this to Sami as a feature request for 
duktape.

I have been working on this over the weekend.  I hope we can get through 
it and support amazon.  Do we know any gangs of CS undergrads with itchy 
fingers for an interesting problem?

I'm erring on the side of pessimism given the conflicting reports and 
assuming that Karl was able to do something else but not actually get over 
the moat yet.  Does that seem right, KD?

K

[Edbrowse-dev] Continuing Amazon issue

[Karl Dahlke]

Well I'm rather baffled. I can sign in.
I got rid of all the amazon cookies from my jar.
Used ed to do this so there would be no collisions.
Now I'm starting clean.
I called up amazon.com, clicked on the sign in link, which takes me to the sign in page,
entered login and password, click go, and there I am logged in.
Actually it's Wendy's account, my wife, so you'll see the name Wendy.
db3 log is at
http://www.eklhad.net/amazon.wendy
Yes I get the same uncallable js errors you get, and that is a concern, we have to figure that out,
but it doesn't stop me from signing in.

Karl Dahlke

Re: [Edbrowse-dev] Continuing Amazon issue

[Kevin Carhart]

Wow!  That's excellent.  It just digests fwcim and processes it like it is 
supposed to.  I wonder what's different.  Are you setting your ua to 
something other than edbrowse?

I had no idea we were doing so well on a major site.



On Sun, 7 Jan 2018, Karl Dahlke wrote:

> Well I'm rather baffled. I can sign in.
> I got rid of all the amazon cookies from my jar.
> Used ed to do this so there would be no collisions.
> Now I'm starting clean.
> I called up amazon.com, clicked on the sign in link, which takes me to the sign in page,
> entered login and password, click go, and there I am logged in.
> Actually it's Wendy's account, my wife, so you'll see the name Wendy.
> db3 log is at
> http://www.eklhad.net/amazon.wendy
> Yes I get the same uncallable js errors you get, and that is a concern, we have to figure that out,
> but it doesn't stop me from signing in.
>
> Karl Dahlke
> _______________________________________________
> Edbrowse-dev mailing list
> Edbrowse-dev@lists.the-brannons.com
> http://lists.the-brannons.com/mailman/listinfo/edbrowse-dev
>

--------
Kevin Carhart * 415 225 5306 * The Ten Ninety Nihilists

[Edbrowse-dev] Continuing Amazon issue

[Karl Dahlke]

[-- Attachment #1: Type: text/plain, Size: 792 bytes --]

> Wow! That's excellent. It just digests fwcim and processes it like it is
> supposed to. I wonder what's different. Are you setting your ua to
> something other than edbrowse?

That's a really good question, and the answer is no. I'm edbrowse all the way.
So we have to figure out what's different about me.
I'm linux fedora, actually an old version of the os and just about every library, so you'd think I was the one in trouble.
My curl uses the other library, relative to most of you, I remember that, but hard to imagine why that would matter.
The word amazon never appears in my config file.
Here is the same sign in scenario at db5, maybe you can compare it to yours and see exactly where it diverges.
http://www.eklhad.net/amazon.wendy.5
It's big, 3.5 meg.

Karl Dahlke

[Edbrowse-dev] Continuing Amazon issue

[Karl Dahlke]

Oh I just thought; is it because I don't deminimize? Is our debugger getting in the way?
I had that happen once before, with gdb, but only once in 30 years of programming.

Karl Dahlke

Re: [Edbrowse-dev] Continuing Amazon issue

[Chuck Hallenbeck]

Wow! I need some guidance on using db5 here.

First of all, I retrieved amazon.wendy.5 with wget, which turns out to 
be 3.5 mb. I looked at the first few lines and the last few, and saw 
that the db level was already in effect when "e www.amazon.com" was 
issued, so I set mine in the config file.

I started the script program to capture output, then issued

e  www.amazon.com

I admired all the passing lines, waiting for the chatter to quiet down,
but it never did. There seemed to be timers doing their thing
indefinitely, until I lost patience and exited edbrowse with a q.

What am I missing? Is there another way to capture debug output? I can't 
get to the sign-in screen and do my thing there with all the timers 
chattering at me.

Chuck


-- 
Here In Northeast Ohio also, The Moon is Waning Gibbous (55% of Full)
When your only tool is a hammer, everything looks like a nail.
Sent from Bert's iPhone.

[Edbrowse-dev] Continuing Amazon issue

[Karl Dahlke]

Wow I didn't even think about timers. Oops.

db5
db>/tmp/log
commands

Then look in /tmp/log

I didn't even know timers were running, so that means there is no reproducibility.
How long did I wait to issue the next command?
How many timers fired?
Maybe, for debugging, we should have a way to turn timers off.
But if those timers are important to the functioning of the site...

Karl Dahlke

Re: [Edbrowse-dev] Continuing Amazon issue

[Chuck Hallenbeck]

Okay, I will do that, but first I have been preparing another email to 
the list, let me insert it here:

Hi Karl, Kevin, and all,

Well, I have Karl's db5 run successfully signing into Amazon,  and
I have my own attempt to replicate that data in order to diff the
two outputs.

However, as I indicated, I seem to have a runaway looping situation
in which debugging output never stops long enough for me to issue
other commands following the very first one,

e www.amazon.com

When my patience was exhausted, I typed "q" and pressed enter,
and edbrowse terminated, leaving me with a script slightly smaller
than Karl's.  Here's what I did next, which I hope will be of some use:

Opening Karl's script file, I search backwards for this:  delete cg

and I found this at line 173304:

> delete cg$$134

Turning to my own script file, in which the only two commands I ever
issued were "e www.amazon.com" and "q", I repeated the backward search
and found this at line  116943:

> delete cg$$740

Finally, What I founnd in Karl's file was 523 lines before the end
of the script, and what I found in mine was 524.

It looks to me like my execution is doing something very different from
what Karl's is doing, so I'm not sure what to do next. Should I let
my own effort run longer? Should I post what I have or do something
else with it locally here?  I'm definitely in over my head but will
do whatever it takes to get more useful information.  Trouble is,
I don't know what's useful and what's not.



-- 
Here In Northeast Ohio also, The Moon is Waning Gibbous (53% of Full)
When your only tool is a hammer, everything looks like a nail.
Sent from Lowell's iPhone.