If you click on a country(second time) and the date indicator is on, corona closes about corona HOT 10 CLOSED

I cannot reproduce these issues (which I have seen in a different context, though). The problem is that the program stores the state of the last session.

For this issue, let's start in "virgin" state. In menu "configuration", uncheck "Automatic saving/loading"; delete the cfg file and the data files. After the second start (when cfg has been recreated) the program should behave reproducibly. Play with it, but without data. Hopefully there should be no crashes any more (I fixed a couple of them a minute now).

Then download the data. Play again. When a crash happens, restart the program and repeat the steps you took. If you can reproduce the issue, report it precisely with the steps taken.

Later we can continue with active cfg file. In this case, you should probably post your cfg file; hopefully my app will behave like yours when I copy your cfg over mine.

from corona.

Hi Werner(@wp-xyz),

I cannot reproduce these issues (which I have seen in a different context, though). The problem is that the program stores the state of the last session.

Even with the *.cfg file that I provided?
That is quite strange, and I'll tell you why: I had to rm -rf my corona cloned repo because source/cabout.pas had some hickup and was stuck on either CRLF or LF, so I cloned the repo fresh and compiled it from 3 or 4 commits ago.

It should be as pristine/virgin as you asked. But none the less, let me pull the latest and re-virginate all the things and report back.

Cheers,
Gus

from corona.

Hey Werner(@wp-xyz),

How much are you willing to bet that if you close #43 this one goes away?

Cheers,
Gus

from corona.

Hey Werner(@wp-xyz),

Nevermind, I solved the maps not being compiled but this issue continues.

Cheers,
Gus

from corona.

Did not notice your cfg file. But even when I use it I cannot reproduce the issue. (BTW, I now split the "Automatic load/save" menu item into separate "Automatic load" and "Automatic save" because when the action was checked previously to read the cfg file at program start it would automatically write it at program start and thus overwrite the cfg. Now I can handle your case with enabled auto-loading and disabled auto-saving).

Since you mentioned the map resources: Resource files always have been kind of tricky, and I cannot guarantee that the new map resources are automatically included. Please try a "Run" > "Build" from Lazarus which I know to rewrite the res file.

from corona.

Hey Werner(@wp-xyz),

Something strange is happening on my side for sure.
I've now compiled from the command line, under the source folder, these ways:

1. lazbuild -B corona.lpi; This uses the Default build mode still 💣
2. lazbuild -B --bm=Debug corona.lpi; still 💣 but with loads of more info dump
3. lazbuild -B --bm=Release corona.lpi; still 💣

And from the IDE:

1. Shift+F9(Build) and Ctrl+Shift+F9(Run no Debug); does not 💣 but the Date Indicator is turned off every time I select another country.

This is some voodoo magic here!! The bug only shows when compiled with lazbuild?!?! This is really freaky!!

Cheers,
Gus

from corona.

Hi Werner(@wp-xyz),

EXPLETIVE!! EXPLETIVE!! EXPLETIVE!! EXPLETIVE!! EXPLETIVE!! EXPLETIVE!!

I'm SOOO sorry for wasting your time!!!

Just found out that lazbuild trunk and 2.0.6 have no issue, but lazbuild 2.0.[12,10] are the ones that 💣s
I can reproduce on both inside the IDE.
So it's definitely a problem with FPC 3.2.0!! Progress.... I hope!!

The problem is on line 827 of cMain.pas:

  Chart.ClearSeries;

So whatever you're doing by adding the purple line to the chart, PFC 3.2.0 does not like when you tell Chart to clear the series.

I'm feeling really tired, since I've been up for far too long. I'll debug this tomorrow, or my type of tomorrow: after I've slept enough.

Cheers,
Gus

from corona.

Fixed. A clear indication of poor design of the TChartListBox component (done by myself...): Chart.ClearSeries destroys the series one by one, and each time TChartListbox is notified to rebuild itself. During this process the listbox checks whether the series DateIndicatorLine is nil -- and of course, DateIndicatorLine is soon destroyed but it is not reset to nil...

from corona.

Hi Werner(@wp-xyz),

I'm glad that you found the issue!! And we are human after all, so we are prone to error 😄

What really baffles me is the fact that FPC 3.0.4 and FPC 3.3.1 just gloss over that issue, hein?!

Cheers,
Gus

from corona.

Closed by 624e4c8

from corona.