Checking for new updates puts the service out of order

(seems a bit similar to bug #1233521)

as discussed, looked at the signal and the service is getting confused on multiple CheckForUpdate() call ;)

Hmm, I tried this with s-i 1.8:

Did an update, which worked fine, rebooted. Did another update via the u/i while tailing the log file. Said I was already up-to-date and gave me a Retry button. Multiple clicks on Retry do nothing (and nothing in the log file).

So I kill the system-image-dbus process, edit /etc/system-image/client.ini to bump debugging up to debug level, and try clicking on Retry again, again tailing the log file. Clicking on Retry does nothing - no activity in the log file.

Then I click on Back and on Updates again, and this time I do see that it checks for another update. Of course there isn't one so I again get the "no software update available" message and a Retry button. Clicking Retry produces no activity in the system-image-dbus log file.

*Without* killing system-image-dbus, I hit Back and then Updates again. Now I get activity in the log file and again it accurately tells me (both in the u/i and log file) that I am up-to-date.

So it seems to me that the Retry button is not sending another CheckForUpdate. Is that possible?

One relatively easy fix (although I'm not yet convinced there's a bug in s-i-dbus yet ;) would be to simply exit the process when we find that we're already up-to-date. Let D-Bus activation restart it. The persistence code is currently disabled so exiting should throw away all state and start up in a pristine mode again.

I still think s-i is pretty good about responding to multiple CheckForUpdates, but this would be an easy way to verify.

s/verify/narrow it down/

Thanks Barry, indeed there was a bug that sneaked in a previous commit, on the system settings side, I'm just submitting a fix for that. Once that change is in, the behaviour of the retry button should be restored.

I'm going to test again if I manage to get the panel in a state where it keeps doing the "check for update" with endless spinning then

On Oct 03, 2013, at 03:34 PM, Sebastien Bacher wrote:

>Thanks Barry, indeed there was a bug that sneaked in a previous commit,
>on the system settings side, I'm just submitting a fix for that. Once
>that change is in, the behaviour of the retry button should be restored.
>
>I'm going to test again if I manage to get the panel in a state where it
>keeps doing the "check for update" with endless spinning then

Sounds good, thanks!

This bug was fixed in the package ubuntu-system-settings - 0.1+13.10.20131004-0ubuntu1

---------------
ubuntu-system-settings (0.1+13.10.20131004-0ubuntu1) saucy; urgency=low

  [ Sebastien Bacher ]
  * battery: having 100% power is still "fully charged", even while
    discharging.
  * battery: only display the location caption if the gps settings is
    visible.
  * update: clean small unwanted changes that sneaked in a previous
    commit, that should make the "check for update" button do the
    correct thing again, rather than having the label hidding and the
    button stop working. (LP: #1234703)

  [ Iain Lane ]
  * When manually selecting a timezone, navigate back to the previous
    screen (Time & Date) after one second. Don't highlight all cities
    the first time you've selected a new timezone, just the chosen one.
    (LP: #1232114)
  * Check the passed mount point when measuring directory size, to avoid
    a GLib critical message.
  * [about] In the storage subpanel, check the space for drives Qt
    reports as "Unknown".
  * Tweaks to the default layout - Make the text smaller - Make the text
    wrap at the edge of the bounding box of the grid item - Centre the
    grid horizontally - Dynamically space the grid elements according to
    how many columns there are in that category. (LP: #1223827)

  [ Ubuntu daily release ]
  * Automatic snapshot from revision 429
 -- Ubuntu daily release <email address hidden> Fri, 04 Oct 2013 05:43:12 +0000