win_sparkle_init doesn't trigger update check on first startup (regression)

vslavik / winsparkle

App update framework for Windows, inspired by Sparkle for macOS

http://winsparkle.org

Other

1.31k stars 267 forks source link

win_sparkle_init doesn't trigger update check on first startup (regression) #144

Closed solusipse closed 7 years ago

solusipse commented 7 years ago

Description: In 0.5.3, calling win_sparkle_init() results in performing a silent check. With 0.5.4 this doesn't happen anymore - no window is displayed apart from settings used, no callback functions are called.

Reproducing: Master and latest release (0.5.4) are affected. 0.5.3 seems to work just fine. Can be reproduced with both examples provided in the repository.

Had no time to check which commit breaks it, sorry. Can do that tomorrow (maybe even provide a fix).

vslavik commented 7 years ago

Had no time to check which commit breaks it, sorry. Can do that tomorrow (maybe even provide a fix).

Please do, I'm on vacation and can't. Prime candidates are cb4b0fd7c788bed0b2e94eb19832e05ab32a2916 (#126), f0ddddc274cf94fcfaa0bd4ee7af094fc12e4515 and 2b326cbbcccb940791a617a68d19674542cd0c8c (#141). I'm petty sure I tested it, so my bet would be on the last one.

vslavik commented 7 years ago

I just verified that checking for updates does in fact work correctly.

If you believe otherwise, please provide detailed instructions on how to reproduce it in the example_psdk sample included with WinSparkle, including information about your OS, relevant registry values (such as LastCheckTime) and times of invoking any operations necessary to reproduce.

cloudshao commented 7 years ago

I ran into this issue as well. Repro:

Delete the LastCheckTime registry value
Run example_psdk, don't press the manual check button

Result:

No update dialog appears

Expected:

Update check happens and update dialog appears

Please see my fix in PR #146

vslavik commented 7 years ago

Ah, on the first ever check — OP's use of "anymore" threw me off and I tested only with pre-existing LastCheckTime records (and the subsequent lack of communication despite promises made it look like your usual PEBKAC).

Thanks a lot for looking into it in more detail, pointing out the broken case and providing a PR — I fully agree with your analysis in #146.

cloudshao commented 7 years ago

Cheers! Thanks for all your hard work on winsparkle. 😄

vslavik commented 7 years ago

Just uploaded v0.5.5 with @cloudshao fix.

solusipse commented 7 years ago

Very sorry for my delayed response. When I meant anymore, I meant for the latest version - sorry for not being precise enough. Thank you for fixing that!