Dear Rachel,

I'm sorry to hear that you are experiencing some problems. What you describe sounds like a bug: I will have to investigate more to find out why it happens. To get me started, I have a few questions for you:

1) Is it all right for me to look directly into your review? This is likely to be necessary in order to isolate the issue.

2) Does this always happen with the 'Boundaries' and 'Enhance staff role' codes or does it depend on other circumstances as well? It may depend on the specific document you are coding, but it is more likely to depend on what text you have selected.

3) Is it repeatable? Does it happen each time you try to do the same thing or does the problem come and go? I very much hope it is repeatable, otherwise I might struggle to find the cause.

4) Assuming it's repeatable, would you be able to provide precise details on how to cause the error? This would help me most.

Many thanks and sorry for the inconvenience.

Sergio

Thank you Rachel,

I will now log on your review now and have a closer look, I will be adding some coding for my tests and will make sure I will clear up everything I did before I leave. Will let you know my findings as I go along.

Sergio

Hi Rachel,
Re "should I have logged out?", no, that's fine. EPPI-Reviewer allows for concurrent usage, making it safe for both of us to work on the same review.

Re your original problem. I did plenty of attempts to recreate it, and could trigger the same error in one instance: unfortunately, all my efforts to isolate the causes were unsuccessful. The good news is that I have a very clear idea of what happens when the problem occurs, and therefore I can provide some (preliminary) help.
1) This has nothing to do with what code you are trying to apply: as you have found out, it may occur with different codes. It is completely unrelated to what other codes you have applied to the same text, and it is also not related to what text your want to code. What is happening is that your local application "forgets" some of the information it holds and as a result it sends the wrong command to our server when you click "Code selected Text".
2) You can quickly workaround this issue, whenever it happens. Please navigate to the next or previous item ("Previous / Next" buttons in "citation details"), return back to the item you were working on and re-open the "Text Document" ("Go!" button under "Files", "View text" column). This will be enough to reload the "forgotten" information and allow you to keep working normally until the error appears again.
3) I can easily prevent the error to be raised, but only in a retroactive way: I can change what happens on our server, so that even if the information went missing on the local client (what you see in the browser), the server will not get confused by the inconsistency and make sure that the missing information is re-transmitted back to your side. However, I will not apply this solution straight away. This is because I would be "hiding" the real problem, and make it much harder to understand why the information got lost. In other words, I would be fixing the symptom, but not the cause.
In the next days, I'll try to isolate the cause in a more systematic way, I will not use your review, so feel free to work  normally. If also these attempts will fail, I'll apply the "symptomatic" cure, as it is still better than no cure at all.

Re your second problem:
I have verified that I can see it as described, and I can also re-create it. I have re-coded ("uncode selected text" and "code selected text") for the two genuine passages that you already coded. The resulting report still shows the wrong text, but a clear pattern emerges: the (wrong) text in the report is just before the genuine selection. The good news here is that this does appear to be a bug, but a well-behaved one: I can work on it, and I don't think it will be too difficult to solve it. Please ignore this problem for the next days, it appears to be document-specific and I do expect to be able to fix it soon enough.

I am sorry for the trouble you're experiencing and thank you so much for your patience and collaboration.
Sergio

Hi Rachel,

sorry for the delay.

I have now finished dealing with the second, "well-behaved", bug. The issue happened with some very rare documents (less than one in 1000, according to my rough estimation) that due to their unusual characteristics created the problem you've seen in coded-text reports. This is now fixed and should not happen again in the future.

I still have to identify the cause of the first bug you've experienced: it is an elusive little fellow and I didn't have time enough to hunt for it in a methodical way. Have you experienced the same kind of error in the last few days? After my initial (partial) success, all my other attempts to recreate the issue failed, so I'm starting to think that I will have to opt for the symptomatic cure.

Will keep you updated.

Sergio