I've put in a little time reconstructing the references. I have 80 entries so far, spanning the 1993/1994 boundary, and everything seems to work fine. The performance I'm getting here isn't particularly bad, considering the need to recalculate long runs of year-suffix disambiguation (a few seconds per entry, on an Atom laptop).
I can carry forward in spare time to fill out the whole set, but this angle on the testing doesn't seem to be getting us anywhere.
I'm not sure how much information it will provide for this particular issue, but could you generate a Debug ID that contains the error? Getting a handle on what the processor is actually seeing in input seems to be our best chance of tracking down the failure.
After enabling Debug Output Logging, I started from a correctly disambiguated version of the 230-plus citation piece and added another citation, which, consistent with previous behaviour, resulted in messed-up year disambiguation with suffixes incremented across year boundaries. There were "27888 lines logged", but I got an "invalid response from server" when hitting the "Submit to Zotero Server" button. Please check your mail for the text file.
Here's an observation that might be helpful in isolating the bug. I've got a document (using Zotero 2.x) whose incorrect references ran from:
Smith, J. (1989a, December 8)
to:
Smith, J. (2007cx, August 22).
The year disambiguation suffixes just kept incrementing without ever resetting to "a" for a new year.
Next, I inserted a new citation into this document, which came out as 1999b:
Smith, J. (1998ck, August 6). Smith, J. (1999a, January 4). Smith, J. (1999b, July 25). Smith, J. (2000cm, May 22).
The disambiguation suffixes in the snippet above suggest that the addition of 1999b had the effect of changing what must have been an incorrect 1999cl to a correct 1999a.
Hopefully, you might be able to make inferences from this behaviour.
Next operation in the same document: the insertion of 2000a gets the year 2000 disambiguated correctly but messes up the year 1999 again:
Smith, J. (1998cj, June 25). Smith, J. (1998ck, August 6). Smith, J. (1999cl, January 4). Smith, J. (1999cm, July 25). Smith, J. (2000a, April). Smith, J. (2000b, May 22). Smith, J. (2000c, July 16). Smith, J. (2000d, October). Smith, J. (2002cq, July 22).
Yup, predictable: year 1998 fixed by adding a 1998 citation, all other years messed up again in the process:
Smith, J. (1997cg, September 18). Smith, J. (1998a, February). Smith, J. (1998b, February). Smith, J. (1998c, February 23). Smith, J. (1998d, June 25). Smith, J. (1998e, August 6). Smith, J. (1999cl, January 4). Smith, J. (1999cm, July 25). Smith, J. (2000cn, April).
The key will be it pin down why it's crossing the year boundary in the first place. Once that happens, the data in the registry has been corrupted, and it's hard to draw conclusions from other behavior. I'll write soon by mail with further details.
Upgrade Storage
I can carry forward in spare time to fill out the whole set, but this angle on the testing doesn't seem to be getting us anywhere.
I'm not sure how much information it will provide for this particular issue, but could you generate a Debug ID that contains the error? Getting a handle on what the processor is actually seeing in input seems to be our best chance of tracking down the failure.
Smith, J. (1989a, December 8)
to:
Smith, J. (2007cx, August 22).
The year disambiguation suffixes just kept incrementing without ever resetting to "a" for a new year.
Next, I inserted a new citation into this document, which came out as 1999b:
Smith, J. (1998ck, August 6).
Smith, J. (1999a, January 4).
Smith, J. (1999b, July 25).
Smith, J. (2000cm, May 22).
The disambiguation suffixes in the snippet above suggest that the addition of 1999b had the effect of changing what must have been an incorrect 1999cl to a correct 1999a.
Hopefully, you might be able to make inferences from this behaviour.
Smith, J. (1998cj, June 25).
Smith, J. (1998ck, August 6).
Smith, J. (1999cl, January 4).
Smith, J. (1999cm, July 25).
Smith, J. (2000a, April).
Smith, J. (2000b, May 22).
Smith, J. (2000c, July 16).
Smith, J. (2000d, October).
Smith, J. (2002cq, July 22).
Smith, J. (1997cg, September 18).
Smith, J. (1998a, February).
Smith, J. (1998b, February).
Smith, J. (1998c, February 23).
Smith, J. (1998d, June 25).
Smith, J. (1998e, August 6).
Smith, J. (1999cl, January 4).
Smith, J. (1999cm, July 25).
Smith, J. (2000cn, April).