ajax errors show 'following errors occured' or a big dump of the html source for the oops page

Bug #521447 reported by Andreas Noteng on 2010-02-13
30
This bug affects 6 people
Affects Status Importance Assigned to Milestone
Launchpad itself
High
Unassigned

Bug Description

If the internet connection dies during AJAX submission, or the permission is denied then the client side AJAX flashes up a useless, and uninformative error box with no guidance for the user about what they should /do/ about it. There are side-effects:

  1. The user gets frustrated, puzzled, annoyed (needs reassurance)
  2. There is no method (without an OOPS number) to report to the Launchpad devs
  3. The user loses their input data effort since they can't resubmit it (needs explanation)

The error dialogues are completely opaque to the user, eg.:

    "The following errors were encountered: (,'displayname','launchpad.View')"

where a messages such as:

   "Connection error"
   "Unable to connect to Launchpad"

or a message indicating server failure (with the returned OOPS number) might be preferable. The message also needs to indicate what the user should /do/. Eg. reconnect and resubmit, report the bug (with direct link to malone/+filebug with subject already filed out as OOPS).

In the case of debugging bug #634847 the inability to get detailed or OOPS information back from the users who were seeing it cost a lot of research time. If the client side Javascript receives an OOPS or other fatal error from the server during an AJAX submission it should deliver this information to the user so that it can be reported.

At a minimum the OOPS should be displayed to the user with the same guidance as the main website error box ("please report this to...").

Andreas Noteng (andreas-noteng) wrote :
description: updated
Curtis Hovey (sinzui) on 2010-02-13
affects: launchpad → lazr-js
Changed in lazr-js:
importance: Undecided → Low
status: New → Triaged
Deryck Hodge (deryck) wrote :

I'm adding a malone task to this bug, since we have a general problem with the AJAX widget error reporting on malone. Bugs continue to come up from time to time about this. My thinking is that this will be a general "better error reporting" bug, and once we have the infrastructure in lazr-js, we can open bugs about any remaining specific cases after this work is done.

Changed in malone:
status: New → Triaged
importance: Undecided → Low
tags: added: error-handling errors javascript ui
Deryck Hodge (deryck) wrote :

This is continually a problem, so I'm raising the importance of this.

Paul, do you want to dupe bug 408491 against this one, just to have a unified bug and also to raise the heat on this issue. ;)

tags: added: dhrb
Changed in malone:
importance: Low → High
Changed in lazr-js:
importance: Low → High
Deryck Hodge (deryck) wrote :

Also, bug 638373 notes that we need to provide OOPS numbers in any error we provide, with a note about who to contact for help, mcuh like we do in normal error pages.

Paul Sladen (sladen) on 2010-09-15
summary: - Uninformative error message
+ Uninformative AJAX error messages: provide OOPS and "next step" pointer
Paul Sladen (sladen) on 2010-09-15
description: updated
description: updated
Deryck Hodge (deryck) on 2010-11-30
summary: - Uninformative AJAX error messages: provide OOPS and "next step" pointer
+ Bad AJAX error messages: provide OOPS and "next step" pointer
summary: - Bad AJAX error messages: provide OOPS and "next step" pointer
+ ajax errors show 'following errors occured' or a big dump of the html
+ source for the oops page
Robert Collins (lifeless) wrote :

I know at least some of the time we show ajax oops ids now; not show how widespread or robust it is yet.

Deryck Hodge (deryck) on 2011-07-07
tags: removed: dhrb
Robert Collins (lifeless) wrote :

Fix for this will be in-lp itself.

Changed in lazr-js:
status: Triaged → Invalid
no longer affects: lazr-js
Robert Collins (lifeless) wrote :

I'm going to narrow the scope on this bug on 'internet connection lost' cases. For that we probably want a gmail like 'retrying now' - if we can detect that.

Robert Collins (lifeless) wrote :

However, further checking shows no disconnection symptoms here, so just going to dupe on the earlier bug.

To post a comment you must log in.
This report contains Public information  Edit
Everyone can see this information.

Other bug subscribers