[MLZ] variant fields behaving strangely

Sarala

I'm trying to add a Pinyin variant to a Chinese title. The interface is acting buggy and unpredictable. I can't get it to save my edits. When it first creates the Pinyin field, it automatically fills it with characters. When I replace the characters with pinyin and tab to the next field, it reverts to characters. Finally I tried pasting in the pinyin between two of the characters, then deleting characters one by one, tabbing to the next field after each action. It worked until I deleted the last remaining character -- then it filled them all back inǃ It's really strange. Does anyone know what's happening?

Longtime Zotero user here, just starting with MLZ. Earlier today I successfully added an English field and several pinyin fields to this same record. The buggy behavior seemed to come out of nowhere, unless it was the fact that the title has two variants attached (en and pny) while the others just have pny.

fbennett

Sarala,

I haven't been able to reproduce that (yet). Could you export the data for the item as Bibliontology RDF, paste it to http://gist.github.com, save as a Public Gist, and then post the URL back here? With the same item data, I may have better luck reproducing the fault.

fbennett

Actually, there is something you can test locally before that. Export the item (as above), then re-import it, and check to see if the newly imported item shows the same buggy behaviour.

If it does fix it, don't delete the buggy item; there are some tests we can run on it, to figure out why it failed. If it doesn't fix it, post the data and I'll take a look.

fbennett

Sarala,

Good news for you, I think. I was able to reproduce the error (and a couple of others). A fresh version is up (4.0.26m512), which should be much more well-behaved.

Post again if you have any more problems; although we have had bugs in the early running, the new interface is a better piece of software, and it should settle down quickly.

Sarala

Frank,

Thanks for the quick response. I didn't even get around to doing those tests until after you'd released the new version. The good news is, exporting and re-importing resulted in a record that is editable (though all the MLZ variant fields had been removed). I've also been able to edit other records without problems in the new version.

The original buggy record is still buggy under the new version of MLZ, so I've created the gist anyway. But perhaps whatever corrupted it in the first place has been fixed?

https://gist.github.com/saralakumari/0e3adf74018b08b8ad4a

fbennett

That is very likely. In the bug you encountered, MLZ was saving a record with the same language marked in the main field as in one of the variants. Only one of the can be displayed by the logic of the user interface, so the phantom record lurked and interfered with oprperations in surprising ways.

You can reveal the phantom record by adding some arbitrary language in Preferences, and the switching the headline field to that language.

Apart from that one issue, the record should be okay.

If you run into fresh problems, give a shout.