Field contents copied when finding/selecting "not selected" recorcord

This has happened many many times over the past several years working on this particular database, but it is rather rare and so just worked around it and wondered if it was user error or something. But after happening twice today, going to try my luck reporting it here.

PanoramaX 10.2.2b1, Mac M1 Pro Sequoia 15.3
(but has been issue since maybe 2020?)

Database with about 4000 records, with custom form open and
standard view table open. I often work on selected records, but then need to compare info with an unselected record, so click on blue “visible/total” box, or Cmd F, and use “New Selection” search dialog box, double-click on record needed, chose “Yes” from “Clicked record is not selected, select it?” popup.

The newly selected record (data which may not have been changed in years) then has complete text field from the text field of the completely different record I was just editing, replacing its own text field. This happens instantaneously, no user input, I click “Yes”, look at the form, new text I was just editing is now in same field in old record.

Cmd-Z does not fix. Quit w/o saving saves the old (unselected) record, but doesn’t seem to save current editing in working record in selected set (I think this is correct, haven’t track all details closely).

Is this a cache when using a form thing? I don’t know if I save before searching for record. If needed I can try to follow some procedures if not easily reproducible, but this has happened at least 100 times over the years.

Thanks

Same thing happened again just now. Since I’m always double-clicking a single record from the search box, maybe the double click triggers something to copy info from one record into the other? Not always the same field, but only one field gets copied.

I did my best to reproduce the steps you described, but it worked fine for me. So probably I am not quite understanding all of the necessary steps, or there is something about your database design that is different from mine.

Thanks so much for taking time to investigate, appreciate it. This behavior has been sporadic enough that there must be some minor conditions that I haven’t noticed yet, so I will try to monitor specifics more closely. If there are any procedures to help with tracking this, logs or debugging mode to turn on, or things I could do to help, let me know. Here are a few more details—

While possible something quirky with this particular database, I will say it is super plain vanilla, out of the box: 21 fields, 1 time-stamp procedure, 2 forms, really nothing complicated.

I have also noticed other anomalies when doing this double-click selection procedure:

Random recorded added. Occasionally, also rather sporadic, instead of the record that is double-clicked, another random record, not one listed in the search box (so not a clicking error), will be added to the selection list and selected. If I then repeat the find, double-click process, it will work and the correct record will be added.

Field edits not saved disappear. I edit a text field in Form_A, not doing Save, then go do the same process (find and add an unselected record), now this new record is selected and visible in Form_A. I return to record I was editing, and all added edits have disappeared. This seems like a straight-up bug, happens not every time, but easily reproducible. Just saw it happen several times now and this last time, to see if could recover text, did Cmd-Z, and after another Cmd-Z, PanoramaX totally crashed. Crash report saved if wanted.

It just seems like there is something not right about that particular find box double-click selection process.

Crash reports are automatically sent to us, so you don’t need to send them manually. And the automatically sent reports have additional symbol information that your copy of the report doesn’t include.

The scenario’s you are describing sound like situations that have been throroughly tested. So I’m stumped for now.

Now occurs to me, maybe something to do with this? Form_A has five large WebBrowserObject windows that constantly pull in whatever URLs are in five text fields, these also tested and well-behaved? I guess I can make another copy and pull these off the form and see if it makes a difference.

I don’t know of a problem with Web Browser Objects, but I think your idea of testing with those removed is definitely worth trying.

I removed WebBrowserObject windows from the form and had the same result of previous selected record text getting copied to newly select record, followed by a crash, so you should have received crash reports on that. Thinking that maybe just removing the fields might not have cleaned the bad stuff out of the form entirely, I then removed the Form_A completely and worked with text field only Form_B.

This time doing the same routine, the newly selected record appeared, there was a warning dialog box showing the full text of the previous record, asking "Integer values can only contain +/- followed by one or more numeric digits. “Re-edit Input” “Cancel Input.” Had not seen that before. Returning to original record, there was no added text in the field. As a plus, doing a few Cmd-Zs to see where that got me did not crash immediately as before with WebBrowserObject form. However doing further simple typing in field then Cmd-Z did not undo typing at all, so basic editing operations were somehow kaputt, PanX needed a restart.

For me it seems like the newly added text in the first selected record is not properly added to that record, by virtue of the fact that on returning to that record, it has disappeared, and there is a crash when you ask to undo that typing. Or other thing that happens is, well, woops, didn’t write it in that record, I’ll stick it in the newly selected record, or I’ll put the whole text in an error box and complain about integer values. How can you leave a record, go to next selection via the Find Unselected, then go back to first record and it has erased new edits from 5 seconds ago?

For both of these tests, I did screenshots before each action, so can send those over to see the evidence, with some description of what I was doing, if this is worth pursuing, can also send database, all public info in it anyway. Or export data and just start over and hope for the best and this never happened?

If you want to send the database, I’ll look at it. But please, send very specific instructions for how to reproduce the problem.

Before you do that, does this database have any implicitly triggered procedures, especially .CurrentRecord or .ModifyRecord?

I’m adding more instructions to the screenshots, am going to run it a few more times to verify, want to respect your time if you look at this.

Only one procedure, and going to assume this .ModifyRecord is safe:

TimeStampLastModifiedDisplay=superdatepattern(TimeStampLastModified,"dd Mon yyyy @ ",“hh:mm”)

I would try removing that and see if that fixes the problem.

If you are displaying the time stamp in a form, I would just put the formula in a Text Display object, rather than keeping a separate field for the text version of that. I guess if you think you need to see the time stamp in the data sheet then this .ModifyRecord would be the only way.

Would have like this to fix things, deleted .ModifyRecord procedure and all related fields and form objects, no luck.

Incoming database and detailed info with screenshots on what is happening, at least for me, hope it is something simple or something stupid that I’m doing.

I wonder if problems I am seeing are related to these reports below? (But was also happening before Sequoia.) I almost always access Find/Select on toolbar. But just tried only using Cmd-F or Menu selection, did work sometimes, not always, as edits disappeared and at some point using Undo and Redo Edit Data Cell, nice text in original cell turned into this mess, weird behavior in a simple text database.

No, they aren’t related. The report you mentioned was a specific problem related to Sequoia, and occurred every time. A fixed version was released within a week or two.