Gramps version: 5.2.1 on Windows 10, sqlite backed database.
I’m surprised to see the surname “Person”, when opening it is all light green, as if this is some strange copy or anonymous version of persons. What could have caused this, and is it safe to remove these "Person"s?
These persons have a tag with a green color.
The thing is, I can’t remember adding these persons. Please note the contents of these persons; the names of these persons are actually “0001, Persoon”, “0002, Persoon” etc. etc. And the contents are pretty bogus, no relations, no parents, no children, the gender of all these persons is “unknown” and there are a few thousand (over 5000) of these persons in the database. So it looks like some strange bug - maybe a result of an earlier import, but I really can’t think of a good reason (I never import that many persons at once from another source).
Would there be good ways to analyse the source of these persons or creation date or something like that?
You can configure the person view to add the date of the last change, so that it looks like this:
The configuration button sits next to the ones that switch between flat and tree view.
You could also look at a backup, by creating a new tree, and importing a backup file into that.
Have you used another program before Gramps? Have you ever tried a CSV import?
Thanks. The last date was very recent. It must have been a recent used tool, maybe the “comparison of duplicates” which I might have interrupted with a hard-break by closing the Gramps applications from windows - might have been an unresponsive tool or something.
Have deleted all these empty persons now.
This looks like the output from the Date Parser Display Test tool. In addition to people, the tool also creates many birth and death events and the Pass and Fail tags which are coloured green and red.
You should have got the warning “This test will create many persons and events in the current database. Do you really want to run this test?” together with an opportunity to cancel the tool.
I ran that here, and it creates thousands of such persons indeed, with surname ‘Persoon’, unknown sex, and a green flag that says "Passed’, in Dutch.
But then the next question is: How did @Hace run this in Windows? The debug menu’s aren’t normally visible in a release version of Gramps.
Well, a few weeks ago I was surprised to see over 50 plugins that I could install. I just might have installed the Date Parser Display Test tool indeed, not sure.
This is what I can see in one of the menus: “Check localized Dat Displayer and Parser”
Which indeed comes with this warning:
Which says it will add a lot of persons and events. I guess I did not notice that before, and have probably run that accidentilly.
I have no clue how that plugin got added to this version of gramps on windows… but as said I do remember that I once saw a plethora of possible plugins to install and probably installed it at that time.
I have, and it’s our fault, which is visible in 5.2.3 for Windows too:
As you can see here, I was able to run this tool here too, and that’s because this 5.2.3 is actually a developers version. And you can see that by looking at the version number, which is not just 5.2.3, like it should, but has some extra text behind it.
If you select About Gramps in 5.1.6, you will see a version number like 5.1.6-1, and in that, you can not download these add-ons in the usual way, even though they do exist, and there is no ‘Foutopsporing’ menu either.
Ok, so you have released a dev version as a production release?
This is what I see in my 5.2.1 version:
Do you think the windows executable over here is thus a dev version instead of a release version then?
Commit aa03f5a contains a change to the AIO build script that downgrades cx_Freeze to 6.15.9. Everything else is the same as the standard 5.2.3 release. At the moment there is no way to specify r2 or to suppress the commit reference.
There is no separate developer version. We just run Gramps without the optimisation -O flag.
Does that mean that, that -O flag should be added to the shortcuts made by the Windows installler?
I’m asking, because I did a check with the 5.1.6 AIO from GitHub, and in that version, there is no debug menu, and the version number is displayed like it should, as AIO 5.1.6-1.
Yes, that’s right. A normal release has a normal version number, like 5.2.1, or 5.1.6-1, and it does not show the ‘Foutopsporing’ menu.
You can see this yourself, if you install 5.1.6 from here:
But I don’t recommend that on a system where you already run 5.2.x. I tested this, to make sure that the old version was OK, and I have screenshots to prove that.
Thanks for all answers. I’ve tried executing with the -O flag but unfortunately the windows-startup-icon does not pickup the flag for some reason; when adding the flag the program is not starting at all. For now, I’ll just ignore the added debug menu and will soon upgrade to another version. My recommendation for an upcoming windows AIO version would be that by default that additional -O flag would not be needed to remove the that developer/debug menu (that is, if that was the original reason it appeared).
Thanks.
Yes. The -O flag sets the __debug__ variable to False. When set to True we enable certain developer features.
Note: Updated to correct and clarify.
And that suggests that there is something missing in the AIO installer that we have since 5.2, because the old one did suppress that, and the Linux installer does the same.
The Windows installer has commands in the nsi template that overwrite the VERSION variable in the version.py file. This always includes the git commit hash. I’m not sure why they want to do that.
Also the code for the revision number doesn’t work in our current GitHub action. Fixes are always welcome.
The code submitted in the AIO pull request for v5.2 is publicly available. I’m not sure where to find the code used for v5.1.
We probably still need a Windows maintainer to take control of this and test releases prior to uploading the release asset.
Me neither, and to tell you the truth, I have no idea who ‘they’ are. And I think that we must make sure that there is some consistency in our releases, which means that for an AIO release, the appbuild variable must be set to a simple number, just like it is for the other platforms, and like it was for 5.1.6-1, also on Windows.
Another thing is, that when I install 5.1.6-1 on Windows, I don’t see that -O on the command line for grampsw.exe, and no debug menu either, so there must be another path that suppresses that menu.
I plan to find time for more testing later this week, but don’t volunteer to become a Windows maintainer, because it’s not my main platform. I can help with testing though.
The pull request was submitted by Jean Michault, but the authors in the file are Josip (pisoj) and Paul Culley.
My guess is that the original script worked well, but was not intended to be run as a GitHub action. I also agree that the “branding” is not a good idea.
This topic was automatically closed 30 days after the last reply. New replies are no longer allowed.
