App stays and not quit, when we deal with "Check Automatically". about winsparkle

[kenjiuno]

Hi.

I have posted a pull request (#45) to fix this issue, though I'm not sure if it is good solution.

Thanks
kenji uno

[kenjiuno]

Hi.

If you try to show CheckUpdate dialog again in example_psdk.exe:
Delete [CheckForUpdates] value from the registry key [HKEY_CURRENT_USER\Software\winsparkle.org\WinSparkle Example App\WinSparkle].

[vslavik]

Did you try with the very latest sources (it's always worth doing)? Specifically, with sources that include 8a440b6? I'm almost sure this is a duplicate of #42 and was just fixed 3 days ago. Could you please check if that's the case?

[kenjiuno]

Hi.

Did you try with the very latest sources

Yes, I have tried and worked on latest master 2ee16f6.

The fix 8a440b6 is included and it didn't fix my problem.

My issue is about "dialog's parent window":

• [Window2] owns [Window3], because wxDIALOG_NO_PARENT isn't specified. I don't know whether this is specification/bug of wxWidgets or not. You can check it by using Spy++.
• If you close [Window2] by selecting one of [Btn2][Btn3], [Window3] is also destroyed. It seems that it is done without required post processing of wxWidgets. It means that wxWidgets treats [Window3] is still alive, so app won't quit, because UI thread cannot be terminated.

ps: I'm using winsparkle on company's product and it works very well, thanks!

[vslavik]

My issue is about "dialog's parent window":

(Your issue was with application not terminating. Nowhere in the description was it apparent that it could be related to parent-child relationship, so the assumption that it was related to 8a440b6 was IMHO a reasonable one.)

Thanks for the follow-up explanation! This parentship relationship between the windows would be correct without wxDIALOG_NO_PARENT if [Window2] was created first; it's highly surprising to me that wx would reparent later, when the dialog is already being shown modally. That's a very valuable observation and makes the reason for #45 much cleaner, thanks a lot for that. (It's exactly the kind of description that would be great to have in the commit message itself, too.)

I still don't understand how could 8a440b6 not work, but that's a problem with me, not this issue and PR — I must still be missing something :(

[kenjiuno]

Hi.

Sorry, I have found a side effect brought by this patch!

[kenjiuno]

Hi.

My mention "a side effect" at the prior post is by my misunderstanding. Sorry!