Warning -- cell XXX no longer exists

[Stan Schymanski]

Dear all,

Since upgrading to 4.7.1., when I evaluate cells in a worksheet one by
one, then delete one that I already evaluated, things become awkward
and in the end when saving, closing and re-opening, that cell is there
again! In the terminal window, I see "Warning -- cell 457 no longer
exists", probably for each cell I deleted, not sure. This does not
happen in sage 4.7. Is this a known issue?

Cheers
Stan

[Stan Schymanski]

CORRECTION: It does happen in sage 4.7. for this particular worksheet.
The worksheet seems corrupted (see my previous post about inconsistent
worksheet names). If anyone has an idea how to repair it, please let
me know. In the mean time, I will try to save it to a file, delete it
from my notebook and then re-import it. Sorry about the noise.

Cheers
Stan

[kcrisman]

Stan, you may want to look at "Edit" in your worksheet as well;
sometimes that will show a more "stable" version of what is happening.

[Stan Schymanski]

Thanks, but edit takes me back to the problem "Bad Request Maximum
length of 102400 bytes exceeded.", so I can't change anything.
Anyway, exporting, deleting and re-importing solved this problem and
the one with the worksheet name I mentioned earlier.

[P Purkayastha]

This commit is probably a fix to your problem:

http://code.google.com/r/rkirov-flask/source/detail?r=5cac8e326cf71ebedc98dc59456dc2194576ecbd#

[Stan Schymanski]

The problem just re-occurred, plus my history has been emptied, so I
just lost several hours of work. Purkayastha, could you let me know
how to apply your patch and what it does? Another symptom is that upon
re-opening of the worksheet, I get into a loop of "Read localhost" and
"Waiting for localhost" while Jsmath keeps circulating between
processing math of 0% and 2%. The most annoying thing is that my
previous revisions are lost, but I don't know if this is related.

On Sep 24, 6:27 am, P Purkayastha <ppu...@gmail.com> wrote:
> This commit is probably a fix to your problem:http://code.google.com/r/rkirov-flask/source/detail?r=5cac8e326cf71eb...

[Stan Schymanski]

By the way, when I shutdown the notebook server, Jsmath finishes
processing without problems, so it does not seem to be a Jsmath issue.
Any help with troubleshooting and fixing this issue would be great.
Thanks!

[P Purkayastha]

Hi,

  I am not sure why, but I seemed to have assumed last time that you were using the flask notebook. Probably because this problem was seen only with the flask notebook and not the normal notebook. 

  Since you are having so many issues, perhaps you can try the flask notebook and see if you have better stability? The instructions for installing it is in http://code.google.com/r/rkirov-flask

  There is some other problem which is I think hard to reproduce that is present also in the flask notebook. If you have a notebook open for a long time (several hours) without doing anything with it, then it tends to lose connection to the webserver. At that point, whatever you write in the cells of the notebook is essentially only there inside the browser and is not communicated back to Sage. If you "evaluate" something, it will simply appear "hung" and never finish. Is this the kind of errors you are seeing? Since you mention "several hours of work" is that the time during which you were idle and not communicating at all with the server (like not even a "Save" or an evaluation)?

[Stan Schymanski]

I think that I have run into the lost connection after long idle time
before, with the normal notebook, or whatever notebook came with sage
4.x. However, this was not the problem here, as I was able to evaluate
cells. The problem only occurred when I navigated away from the site,
either by saving and closing or by saving and then clicking on "undo"
to check whether the revision was saved. There is a thread at sage-
support, that describes it a bit better and also points to a way
around it:
http://groups.google.com/group/sage-support/browse_thread/thread/98f9c4098fa9a07f
Basically, it seems that Firefox stores a copy of the notebook
somewhere and does not update it. Upon re-opening it loads an old copy
and overwrites the changes when I re-evaluate the cells. The way
around it was to press "Reload" in Firefox after re-opening the
notebook. Very strange.

On Oct 4, 7:13 pm, P Purkayastha <ppu...@gmail.com> wrote:
> Hi,
>   I am not sure why, but I seemed to have assumed last time that you were
> using the flask notebook. Probably because this problem was seen only with
> the flask notebook and not the normal notebook.
>
>   Since you are having so many issues, perhaps you can try the flask
> notebook and see if you have better stability? The instructions for

> installing it is inhttp://code.google.com/r/rkirov-flask

[Stan Schymanski]

By the way, I do frequently leave the notebook open and only resume
work after a few hours. Would everything done since the last save be
lost if the connection was lost in the mean time? That would
definitely be a show stopper for me. All of this would not be a
problem if we had not removed the automatic saving of snapshots as a
result of this thread:
http://groups.google.com/group/sage-support/browse_thread/thread/4b20f36ea11a8f77/ce75e98188bbc994?lnk=gst&q=schymanski+save#ce75e98188bbc994
In that thread, I suggested to keep automatic saving of snapshots but
only keep the last 30 or so snapshots. Unfortunately, my wish was not
considered and automatic snapshot saving was removed. I had never lost
any work before then but lots of hours of work since then.

Stan

On Oct 4, 7:13 pm, P Purkayastha <ppu...@gmail.com> wrote:

> Hi,
>   I am not sure why, but I seemed to have assumed last time that you were
> using the flask notebook. Probably because this problem was seen only with
> the flask notebook and not the normal notebook.
>
>   Since you are having so many issues, perhaps you can try the flask
> notebook and see if you have better stability? The instructions for

> installing it is inhttp://code.google.com/r/rkirov-flask

[Jason Grout]

On 10/5/11 2:27 AM, Stan Schymanski wrote:
> Would everything done since the last save be
> lost if the connection was lost in the mean time?

If I recall correctly, every time you evaluate a cell, that cell is
"saved". The "Save" button just additionally saves any changes you've
made to cells since you evaluated them. I think that is how it works.

Thanks,

Jason

[Stan Schymanski]

Ah, thanks, that's fine then! I noticed that changes made to cells
after evaluation are lost when closing and reopening a document,
therefore I always evaluate a cell after making changes.
I think that the "Save" button additionally saves a snapshot of the
whole worksheet, which is not designed to be overwritten, right? What
I don't understand yet is what happens to the "undo" information of
each cell. During a session, I can change a cell, evaluate it, change
again, evaluate again and then progressively undo my changes by
placing the cursor in the cell and pressing CTRL+Z repeatedly.
However, after closing and re-opening the worksheet, this is not
possible any more. It seems that this could be possible, as the
information is still stored in the Log (link at the top right of the
page). It would be nice to include a description of the save,
snapshot, Log, and undo behaviours in the documentation.

Cheers
Stan

[Jason Grout]

On 10/5/11 4:27 AM, Stan Schymanski wrote:
> Ah, thanks, that's fine then! I noticed that changes made to cells
> after evaluation are lost when closing and reopening a document,
> therefore I always evaluate a cell after making changes.
> I think that the "Save" button additionally saves a snapshot of the
> whole worksheet, which is not designed to be overwritten, right? What
> I don't understand yet is what happens to the "undo" information of
> each cell. During a session, I can change a cell, evaluate it, change
> again, evaluate again and then progressively undo my changes by
> placing the cursor in the cell and pressing CTRL+Z repeatedly.
> However, after closing and re-opening the worksheet, this is not
> possible any more. It seems that this could be possible, as the
> information is still stored in the Log (link at the top right of the
> page). It would be nice to include a description of the save,
> snapshot, Log, and undo behaviours in the documentation.

I think the ctrl-z behavior is just a feature of the browser, not of
Sage. The browser provides a minimal undo for textboxes. Of course,
when you refresh or close/reopen the worksheet, the browser doesn't
carry that undo info forward.

Thanks,

Jason

[P Purkayastha]

Indeed it is the browser doing the undo for you, not Sage. :) 

[Stan Schymanski]

Hooray for Firefox for this feature!
The other feature of coming up with old versions of the worksheet when
I re-open it is quite dangerous, though. Any problems related to this
would probably be avoided if sage AUTOMATICALLY saved a snapshot of
all running worksheets when closing the notebook. Or, in case of a
crash, it could just save a snapshot automatically every couple of
minutes and overwrite it each time to avoid storing too many of them.
It could be labeled as automatic snapshot and then the user would have
no problems distinguishing it from the permanent snapshots generated
when pressing "Save". This would have helped me a lot! Actually, the
last snapshot before closing a worksheet or crashing the computer
should be a permanent one, otherwise it would be overwritten a minute
after re-opening the worksheet again, possibly before the user
realizes that his browser is using an old version of the worksheet.
Having automatic permanent snapshots from the end of each session
would also allow easy reverting to a previous version in case
something goes haywire. Sorry if this is off-topic, but if you agree
that this could be useful, please let me know where to put this
feature request.

Cheers
Stan

[Jason Grout]

On 10/7/11 6:23 AM, Stan Schymanski wrote:
> Sorry if this is off-topic, but if you agree
> that this could be useful, please let me know where to put this
> feature request.

On the sagenb issue tracker: http://code.google.com/p/sagenb/issues/list

Thanks,

Jason

[Stan Schymanski]

Thanks, this is now #58.

Stan

[Stan Schymanski]

I am getting some new error messages connected with disappearing cells
in a worksheet. Does the below mean anything to anyone? This is in
sage 4.7.1 in Ubuntu 11.04.

fixme:exec:SHELL_execute flags ignored: 0x00000100
fixme:heap:HeapSetInformation (nil) 1 (nil) 0
fixme:system:SetProcessDPIAware stub!
fixme:process:RegisterApplicationRestart (L"/Recover",12)
fixme:richedit:ME_HandleMessage EM_SETEDITSTYLE: stub
fixme:richedit:ME_HandleMessage EM_SETEDITSTYLE: stub
fixme:resource:GetGuiResources (0xffffffff,1): stub
fixme:richedit:ME_HandleMessage EM_SETEDITSTYLE: stub
[ConvConfHandler] isPreferred contentType: application/pdf
fixme:exec:SHELL_execute flags ignored: 0x00000100
fixme:heap:HeapSetInformation (nil) 1 (nil) 0
fixme:system:SetProcessDPIAware stub!
fixme:richedit:ME_HandleMessage EM_SETEDITSTYLE: stub
*** NSPlugin Viewer *** WARNING: unhandled variable 18 (<unknown
variable>) in NPN_GetValue()
*** NSPlugin Viewer *** WARNING: unhandled variable 18 (<unknown
variable>) in NPN_GetValue()
*** NSPlugin Viewer *** WARNING: unhandled variable 18 (<unknown
variable>) in NPN_GetValue()
*** NSPlugin Wrapper *** WARNING: unhandled variable 18 in
NPP_GetValue()
*** NSPlugin Viewer *** WARNING: unhandled variable 18 (<unknown
variable>) in NPN_GetValue()
*** NSPlugin Viewer *** WARNING: unhandled variable 18 (<unknown
variable>) in NPN_GetValue()

(npviewer.bin:2936): GLib-GObject-CRITICAL **: g_object_unref:
assertion `G_IS_OBJECT (object)' failed
*** NSPlugin Viewer *** WARNING: unhandled variable 18 (<unknown
variable>) in NPN_GetValue()
*** NSPlugin Wrapper *** WARNING: unhandled variable 18 in
NPP_GetValue()

(npviewer.bin:2936): GLib-GObject-CRITICAL **: g_object_unref:
assertion `G_IS_OBJECT (object)' failed


Cheers
Stan