Nobody has mentioned that anything is amiss, so I’ve now pushed the changes to github.
]]>I’ll take a look on my office machine later today. I’ve never had a problem using a chromebook.
]]>Thanks very much, Rod and Mike. I’m reasonably confident that the issue is fixed now, but it would be good with as much testing as possible. David C, if you see this, could you try editing something in Chrome on Windows?
My cookies are probably good now, but I could restore a copy of the bad ones.
If you think that would more accurately reproduce the state of affairs when you were not able to submit, that would be great and much appreciated!
]]>My computer that runs Windows is on the blink, but if it comes back, I’ll try again.
]]>I just successfully edited Sandbox#1501 in my main profile with the defective cookies.
My cookies are probably good now, but I could restore a copy of the bad ones.
]]>Thanks very much for testing, Rod and Mike. Upon further investigation, I believe that this was indeed the same issue as we had for the file uploader, described here. It was already on the TODO list to remove the spam prevention tool completely; I have now done so.
Could you please test again, Rod, with your old cookies? Could you also please test, Mike? Unfortunately I have not been able to reproduce the issue, so cannot test myself.
If anybody notices any odd behaviour, or encounters an error message about the spam filter kicking in, please let me know.
]]>I finally resolved my cookie problem (I think) in a new empty Firefox profile
In the empty profile I can Edit and Submit Sandbox successfully.
When I copied in “cookies.sqlite” from the old profile
I can click Edit but then when I click Submit I get the error message
Sandbox is locked You must have Javascript on, and cookies enabled, to submit this form.
Anonymous has been editing this page for 1 minutes.
Edit the page anyway Cancel
Then I went to about:preferences#privacy
, clicked [Manage Data]
, and deleted all the ncatlab.org
cookies. Which made things work again.
I’m currently running FirefoxNightly. Two things I noticed since I last tried to fix cookies
You have to reload about:preferences#privacy
if you want to see if cookies have been added. Just clicking on [Manage Data]
again won’t reflect changes.
Individual cookies are no longer shown in preferences
nor is their data accessible. From there I can only delete all 3 cookies from ncatlab.org. If I really want to look at individual cookies I have to use the Storage Inspector in the Web Developer.
After clicking on [Manage Data]
you have to click [Remove Selected]
and[Save Changes]
and sometimes, for no reason that I can tell, the [Save Changes]
button is not present.
I must have made some mistakes the last time I tried to clear the ncatlab cookies.
I’m still keeping my bad cookies in case the nLab could be fixed to deal with them.
]]>I still have the same problem. The display in the Network tab starts with
Name | Status | Initiator | Time |
Sandbox (https://ncatlab.org/nlab/save/Sandbox) | 302 | Other | 532 ms |
Sandbox (https://ncatlab.org/nlab/edit/Sandbox) | 302 | /nlab/save/Sandbox | 418 ms |
Sandbox (https://ncatlab.org/nlab/locked/Sandbox) | 200 | Sandbox (https://ncatlab.org/nlab/edit/Sandbox) | 324 ms |
(The parentheticals are mouseover popups.)
]]>yes, I get the message when I submit. I’ve had the problem for quite a while and rather than tracking down a fix I just use Chrome for nLab edits.
]]>Thanks very much, Rod, much appreciated! I’ll look further into it. Just to be certain: you receive that message when you click ’Submit’, not when you wish to start editing? The message that you are getting may indicate an issue with the spam prevention mechanism, akin to that which I addressed for file uploads recently.
]]>in my wedged Firefox when I try to edit Sandbox I still get
]]>You must have Javascript on, and cookies enabled, to submit this form.
Anonymous has been editing this page for 1 minutes.
Mike, Rod, and David, I’ve now made a couple of small tweaks. Could you please test to see if the problem still occurs?
The tweaks are only an attempt to improve the situation, not to fundamentally solve the problem (and I am not even sure that they will even improve things sufficiently to solve the immediate problem). Really the locking mechanism is a bit fragile, and there will always be the possibility for problems. Keeping the locking information in memory rather than in the database would improve things. But it is difficult to imagine a really rigorous way to handle it. One could drop it completely, and rather check to see if there have been any changes when one submits, sending the author to a page where they can see the diff if so, or possibly to the nForum. Wikipedia takes something like this approach, I believe.
]]>Thanks very much for testing, Rod, it provided me with the necessary logs!
]]>(This cross-posted with Rod’s comment). Yes, I believe this (i.e. what I ask about in #13) is what is happening. Here are some nginx logs.
[removed IP address] - - [25/Apr/2018:07:51:29 -0400] "GET /nlab/edit/Sandbox HTTP/1.1" 200 14572 "https://ncatlab.org/nlab/show/Sandbox" "Mozilla/5.0 (Windows NT 10.0; WOW64; rv:61.0) Gecko/20100101 Firefox/61.0" [removed IP address] - - [25/Apr/2018:07:52:52 -0400] "POST /nlab/save/Sandbox HTTP/1.1" 302 103 "https://ncatlab.org/nlab/edit/Sandbox" "Mozilla/5.0 (Windows NT 10.0; WOW64; rv:61.0) Gecko/20100101 Firefox/61.0" [removed IP address] - - [25/Apr/2018:07:52:52 -0400] "GET /nlab/edit/Sandbox HTTP/1.1" 302 105 "https://ncatlab.org/nlab/edit/Sandbox" "Mozilla/5.0 (Windows NT 10.0; WOW64; rv:61.0) Gecko/20100101 Firefox/61.0" [removed IP address] - - [25/Apr/2018:07:52:52 -0400] "GET /nlab/locked/Sandbox HTTP/1.1" 200 9651 "https://ncatlab.org/nlab/edit/Sandbox" "Mozilla/5.0 (Windows NT 10.0; WOW64; rv:61.0) Gecko/20100101 Firefox/61.0"
I believe this is you, Rod! One can see from this that you go to edit the Sandbox (do a GET), then you Submit (do a POST), then ’you’ try to edit the Sandbox (do a GET) again. This should not happen in the correct sequence: the correct sequence is, after the POST to /save, to proceed (do a GET) to /show. I say ’you’ because I assume that you are not manually refreshing the page, it is something your browser is doing under the hood.
That explains the bug, anyway. It is not a race condition, but something causing the edit page to reload before submission has been completed. If this happens, it is ’correct’ that the page will be locked.
So the question is why the page is reloading! I do not have time to look more into it just now, but I will try this evening (European time) to fix it one way or another.
]]>I just tried editing Sandbox.
The first time I hit submit I got
You must have Javascript on, and cookies enabled, to submit this form.
Anonymous has been editing this page for 1 minutes.
without going back to the edit page view.
On subsequent tries it did go back to edit page.
One thing I’ve noticed is that my name isn’t remembered, even after I change it to RodMcGuire
.
nLab always thinks I’m Anymous
.
Is this enough info?
]]>Could you try to observe whether the edit page is reloading after you click submit, Rod?
]]>Strange, but I’m having that problem in Firefox, not Chrome. I suspect my clearing cookies hasn’t fully worked though I have yet to see if creating a new profile does work.
]]>Mike, could you try to reproduce the problem, this time after clicking F12 in Chrome and looking at what happens in the ’Network’ tab when you click ’Submit’? The problem seems very strange, but one thing that might possibly happen is that the edit page gets reloaded in the time that the Submit request is processing; this would cause the behaviour that you are seeing. I am hoping that you will see the reload in the ’Network’ tab. I cannot reproduce the problem on my side, so it is a bit hard for me to check.
]]>Thanks very much, Urs, this is very good to know.
]]>Richard,
just for the record I say that there was some issue with cookies and saving edits a few months back: This happened after some modification that Adeel had made. When I couldn’t save any entries anymore (upon “Submit” I was always told that I “cannot resubmit” or similar) Adeel told me to clear my cookies. Back then that worked for me, and also for some other regulars here.
Now the present issue must be something different, I suppose, but maybe Adeel might remember and see some relation.
]]>Although removing the cookies might possibly slow down the page load slightly after the edit and hence avoid the race condition in some cases, which might explain why David C had some success with this.
]]>I was able to edit the Sandbox just now with Chrome on a Windows machine. I don’t think it has anything to do with cookies, I don’t think they are involved in the locking process. I still suspect a race condition, which would be consistent with the bug only occasionally appearing. I’ll dig into it further when I get the chance.
]]>I just edited my main page Tim Porter using Chrome on my Macbook with no difficulty.
]]>Clearing cookies didn’t help me.
]]>