severe notebook problems - how to troubleshoot?

Stan Schymanski

unread,

Oct 4, 2011, 6:51:22 AM10/4/11

to sage-support

Dear all,

I am having more and more trouble with a worksheet I have been working
on for the past few weeks, to the point that I can't work on it
properly any more, because changes are not saved and upon quitting and
reopening it gets into an infinite "Jsmath processing" loop. Only
restarting the notebook server allows me to work on it again, but then
some of my changes are usually lost. The problem is described in more
detail here:
http://groups.google.com/group/sage-notebook/browse_thread/thread/7e72cb922fac3e12/c907d7a1b2fa7b4b?lnk=gst&q=schymanski#c907d7a1b2fa7b4b

When googling a bit, I found that someone else had a similar problem
and never got a response:
http://groups.google.com/group/sage-support/browse_thread/thread/ee47acd5e31b62f8/ba3fdf8fc965811c?lnk=gst&q=worksheet#ba3fdf8fc965811c

Like in the above thread, I installed sage 4.7 and 4.7.1 on a 64-bit
Ubuntu (11.04), so maybe the problem is linked to this architecture? I
did sage -testall and got the following output:

The following tests failed:
sage -t -force_lib "devel/sage/sage/sandpiles/sandpile.py" # Time
out
sage -t -force_lib "devel/sage/sage/misc/sagedoc.py"
Total time for all tests: 6989.1 seconds

Could someone advice me what else I can do to troubleshoot the problem
further?

Thanks already for your help!

Stan

Jason Grout

unread,

Oct 4, 2011, 6:58:42 AM10/4/11

to sage-s...@googlegroups.com

One of your comments about the error you were getting when you tried to
edit the raw html of the worksheet seemed to indicate (IIRC) that it was
abnormally large. Can you tell us about the size of the worksheet, any
attached data files, etc.? Can you give us an indication of how much
typesetting is involved? Does all of the text appear (without being
typeset)? Can you see where on the page the typesetting stops, and can
you post the relevant formulas here? Does your worksheet work on
test.sagenb.org (which has been upgraded to mathjax, as well as many
other upgrades)? Are there any javascript errors noted in the
javascript console?

Thanks,

Jason

Stan Schymanski

unread,

Oct 4, 2011, 7:34:10 AM10/4/11

to sage-support

Thanks, Jason, for looking into this.

The worksheet has 3595 lines of text when I click on edit, and no
attached files. I am not sure what you mean by typesetting, but I have
not checked the "Typeset" box. I do, however, use quite a few latex
equations in the html cells, and have a lot of plots. Everything works
fine when I restart the server and the problem does not occur at the
same point every time. I just re-opened it again and tried to
reproduce the problem. Everything worked fine until I got to the place
I last edited, which was a series of plot commands. The plot was still
there, but my additions to the code were missing. So I added a couple
of them again (feeling like in Groundhog day), did "Save and Quit", re-
opened the worksheet and was in the "processing math" loop again. No
error message in the terminal where I started the notebook server
from.
Interestingly, I still had the text version of the worksheet open in a
text editor from before and my edits are there!!
I just restarted the notebook server again, opened the same worksheet
and scrolled to the cell where I made the changes previously. The
changes were gone again. Clicked on edit, searched for the cell there,
and the changes were there! Returned to worksheet, clicked on
"Refresh" and lo and behold, the code cell showed the changes made
previously. This is Firefox 7.01 on Ubuntu 11.04 64-bit. It looks like
Firefox has a local copy of the worksheet stored somewhere and when I
evaluate the cells in this local copy, the worksheet is updated with
this old code and things go haywire.

Where do I find the javascript console? Please excuse my ignorance!

Cheers
Stan

Jason Grout

unread,

Oct 4, 2011, 7:48:00 AM10/4/11

to sage-s...@googlegroups.com

See https://developer.mozilla.org/en/Using_the_Web_Console

If you want to send me the worksheet off-list (or post it somewhere if
it's okay to post publicly), I can take a look at it.

Thanks,

Jason

Stan Schymanski

unread,

Oct 4, 2011, 7:54:45 AM10/4/11

to sage-support

Thanks, Jason! I will work on it for a bit more and keep the web
console open to see if there are any error messages when the problem
re-occurs. At the moment, it looks as if the refresh button helped a
lot.

Cheers
Stan

On Oct 4, 1:48 pm, Jason Grout <jason-s...@creativetrax.com> wrote:

>
> Seehttps://developer.mozilla.org/en/Using_the_Web_Console

Stan Schymanski

unread,

Oct 4, 2011, 8:05:57 AM10/4/11

to sage-support

Jason Grout

unread,

Oct 4, 2011, 8:33:54 AM10/4/11

to sage-s...@googlegroups.com

On 10/4/11 7:05 AM, Stan Schymanski wrote:
> Error in parsing value for 'cursor'.

That looks harmless: http://forums.mozillazine.org/viewtopic.php?t=505660

It looks like you may have a large number of cells (at least 600?) as
well as a lot of images that are loading. Is it just an issue with not
waiting long enough for the images to load? Are there error or warning
messages at the end of the log? I notice that you've only pasted the
first 7 seconds or so after opening the worksheet, while everything is
still loading.

Thanks,

Jason

Stan Schymanski

unread,

Oct 4, 2011, 8:46:05 AM10/4/11

to sage-support

Sorry, I didn't include a large enough range. It seems that cell 648
is the culprit, as this one comes up again and again followed by the
same error message. I waited for 5 minutes now and got flooded with
messages, so I needed to shut down the notebook server before I could
copy and paste the error messages. Here is a section:

[14:38:59.933] GET http://localhost:8000/home/admin/129/cells/648/sage0.png?1
[HTTP/1.1 304 Not Modified 6ms]
[14:38:59.965] Use of getAttributeNodeNS() is deprecated. Use

getAttributeNS() instead. @ http://localhost:8000/home/admin/129/

[14:39:00.787] GET http://localhost:8000/javascript/sage/keyboard/ml?_=1317731940207
[HTTP/1.1 200 OK 2ms]
--
[14:39:06.648] GET http://localhost:8000/home/admin/129/alive?_=1317731946547
[HTTP/1.1 200 OK 1ms]
[14:39:06.836] GET http://localhost:8000/home/admin/129/cell_list?_=1317731946733
[HTTP/1.1 200 OK 1ms]
[14:39:07.072] GET http://localhost:8000/css/main.css [HTTP/1.1 200 OK
1ms]
[14:39:07.110] GET http://localhost:8000/javascript/sage/main.js [HTTP/
1.1 200 OK 2ms]
[14:39:07.161] GET http://localhost:8000/javascript/sage/jsmath.js
[HTTP/1.1 200 OK 2ms]
[14:39:08.138] Use of getAttributeNode() is deprecated. Use

getAttribute() instead. @ http://localhost:8000/home/admin/129/

getAttributeNS() instead. @ http://localhost:8000/home/admin/129/

[14:39:11.590] GET http://localhost:8000/home/admin/129/cell_list?_=1317731950658
[HTTP/1.1 200 OK 4ms]
[14:39:12.063] GET http://localhost:8000/css/main.css [HTTP/1.1 200 OK
3ms]
[14:39:12.084] GET http://localhost:8000/javascript/sage/main.js [HTTP/
1.1 200 OK 2ms]
[14:39:12.110] GET http://localhost:8000/javascript/sage/jsmath.js
[HTTP/1.1 200 OK 2ms]
[14:39:13.874] Use of getAttributeNode() is deprecated. Use

getAttribute() instead. @ http://localhost:8000/home/admin/129/

getAttributeNS() instead. @ http://localhost:8000/home/admin/129/

...

Here is the worksheet content of cell 648 and the adjacent cells:

{{{id=681|
P = plot(1/2*x/(x + 1),(x,0,10))
P += plot(1/2.5, (x,0,10), linestyle = '--')
P.axes_labels(['D2/D', 'D_I/D'])
P
///
<html><img src='cell://sage0.png'></html>
}}}

Division of D by 2.5 is equivalent to a rectangular leaf where the
longer side is 4 times as long as the shorter side.

{{{id=655|
eq_h_c_Nu.subs(eq_Nu_free_Inc)
///
h_c == 3/200*(sqrt(25*(Gr*Pr)^(2/3) - 180*(Gr*Pr)^(7/12) +
324*sqrt(Gr*Pr)) + 5*(Gr*Pr)^(1/3) + 18*(Gr*Pr)^(1/4))*k_a/D
}}}

{{{id=648|
vw = 0.5
Ta = 300
Tl = Ta-10
Dmax = 0.1
P = plot(eq_H_lu.subs_expr(h_cu ==
eq_h_c_Nu.rhs().subs(eq_Nu_free_r_Inc).subs(eq_Gr).subs_expr(eq_nu_a,eq_a_a,eq_k_a).subs(cdict1)).subs_expr(Pr
== 0.72, T_a==Ta, T_l == Tl),(D, 0, Dmax), legend_label='$H_{lu}$')
P += plot(0.5*eq_Rll_n.rhs().subs(cdict1).subs(T_a = Ta,T_l = Tl),(D,
0, Dmax),color = 'green', legend_label='$R_{llu}$')
P.axes_labels(['$D_I$ (m)', 'Heat flux upper side (J m$^{-2}$ s
$^{-1}$ )'])
P
///
<html><img src='cell://sage0.png'></html>
}}}

{{{id=652|
vw = 0.5
Ta = 300
Tl = Ta+10
Dmax = 0.1
P = plot(eq_H_lu.subs_expr(h_cu ==
eq_h_c_Nu.rhs().subs(eq_Nu_free_Inc).subs(eq_Gr).subs_expr(eq_nu_a,eq_a_a,eq_k_a).subs(cdict1)).subs_expr(Pr
== 0.72, T_a==Ta, T_l == Tl),(D, 0, Dmax), legend_label='$H_{lu}$')
P += plot(0.5*eq_Rll_n.rhs().subs(cdict1).subs(T_a = Ta,T_l = Tl),(D,
0, Dmax),color = 'green', legend_label='$R_{llu}$')
P.axes_labels(['$D$ (m)', 'Heat flux upper side (J m$^{-2}$ s
$^{-1}$ )'])
P
///
<html><img src='cell://sage0.png'></html>

Jason Grout

unread,

Oct 4, 2011, 9:05:16 AM10/4/11

to sage-s...@googlegroups.com

On 10/4/11 7:46 AM, Stan Schymanski wrote:
> Sorry, I didn't include a large enough range. It seems that cell 648
> is the culprit, as this one comes up again and again followed by the
> same error message. I waited for 5 minutes now and got flooded with
> messages, so I needed to shut down the notebook server before I could
> copy and paste the error messages. Here is a section:

This all looks pretty normal to me, and it looks like all of the png GET
requests are repeated. It looks like it is retrieving the plots for all
the cells, and cell 648 is just the last cell. Is this from you
refreshing the page several times?

Do you have this problem in other browsers? It could be that FF 7 broke
something.

Thanks,

Jason

Stan Schymanski

unread,

Oct 4, 2011, 9:12:59 AM10/4/11

to sage-support

No, I didn't refresh the page at all. It just keeps repeating the png
GET requests. I'll try with Konqueror and report back.

Cheers
Stan

Stan Schymanski

unread,

Oct 4, 2011, 9:57:07 AM10/4/11

to sage-support

OK, the problem does not occur in Konqueror. Does this lead us
somewhere?

Thanks, Stan

Jason Grout

unread,

Oct 4, 2011, 10:12:22 AM10/4/11

to sage-s...@googlegroups.com

On 10/4/11 8:57 AM, Stan Schymanski wrote:
> OK, the problem does not occur in Konqueror. Does this lead us
> somewhere?

Maybe it's a bug in jsmath with Firefox 7? Does the problem only happen
on this worksheet in FF7, or does it happen in other worksheets with
latex equations in FF7?

If it only happens on this worksheet, then it would be interesting to
try to narrow down what in the worksheet is causing the apparent
refreshes or whatever is going on. Maybe you could do sort of a binary
search in the worksheet---delete half of the worksheet and see if the
problem goes away. If it does, then replace with the other half. If
the problem still remains, delete half of the remaining cells. (Do this
in the "Edit" view so it's easy to select half of the worksheet and
delete). I guess this is assuming that you see the problem before
evaluating any cells.

Thanks,

Jason

Stan Schymanski

unread,

Oct 4, 2011, 10:19:09 AM10/4/11

to sage-support

Well, here we get back to another unsolved problem, related to large
worksheets. I can't save changes when I click on "Edit".
But to answer your first question, no, it only seems to happen with
this worksheet and its predecessors.

Cheers
Stan

Jason Grout

unread,

Oct 4, 2011, 10:31:37 AM10/4/11

to sage-s...@googlegroups.com

On 10/4/11 9:19 AM, Stan Schymanski wrote:
> Well, here we get back to another unsolved problem, related to large
> worksheets. I can't save changes when I click on "Edit".

Even if you delete half (or more?) of the worksheet?

Thanks,

Jason

Stan Schymanski

unread,

Oct 4, 2011, 10:55:15 AM10/4/11

to sage-support

Wow, it works! And I discovered another workaround. Ok, I deleted the
last third of the worksheet and I was able to save the changes.
Hooray! After evaluating a few cells, then clicking "Save and Quit",
and reopening the worksheet, I got into the endless loop again.
However, when I clicked on "Stop loading" and then the "Reload"
button, the page loaded correctly.

Thanks a lot!
Stan

Stan Schymanski

unread,

Oct 25, 2011, 4:16:07 AM10/25/11

to sage-support

There is still a problem that Firefox displays a cached version of a
worksheet from back-when after re-opening, and then updates the
worksheet with this version if one clicks on "Evaluate All" before
clicking on "Reload" in Firefox. I've been bitten by this again and
again, loosing quite a few chunks of my work. There is a simple way to
make Firefox always update the cache automatically, which I thought
might be useful for others reading this thread. I took the below from
http://ticket.betterbuilt.com/KB/a26/how-to-force-firefox-to-reload-the-page-and-ignore-cache.aspx:

Type this into the address bar:
about:config
type this into the filter:
browser.cache.check_doc_frequency
change the value to: 1

more info:
3 - only check if it seems outdated
2 - always use cached version
1 - always check for newer version
0 - check for newer version once per session

Reply all

Reply to author

Forward