Re: gnats(web) occasionally submits duplicate PRs
Jo A Wahle <[email protected]> Wed, 6 Sep 2006 16:30:55 -0500
| Newsgroups | gmane.comp.bug-tracking.gnats.general |
|---|---|
| Message-ID | <OF44D9A1D0.A834668D-ON862571E1.0075FED7-862571E1.00763050@usgs.gov> |
I did remove the button bar from the edit and create confirmation pages, and that prevents the bug from causing any problems. If you do decide to eliminate the confirmation pages and put the confirmation notice on whatever page they go to, that will change the whole flow anyway. -- Jo Wahle, Software Engineer, Engineering Support, SAIC (Science Applications International Corporation ? www.saic.com) Contractor to U.S. Geological Survey ? EROS, Sioux Falls SD 57198 (Earth Resources Observation and Science ? eros.usgs.gov) Voice: 605-594-2514, FAX: 605-594-6940, E-mail: [email protected] "Imagination is more important than knowledge, for knowledge is limited while imagination embraces the entire world." -- Albert Einstein Yngve Svendsen <[email protected]> Sent by: [email protected] 09/06/2006 04:27 PM To Chuck Alexander <[email protected]>, Jo A Wahle <[email protected]> cc [email protected] Subject Re: gnats(web) occasionally submits duplicate PRs Chuck, Jo, thanks for diagnosing this. The return_url system is complicated and the cause of several instances of mysterious behaviour. The mechanisms for displaying confirmations, with subsequent redirects, should be reengineered at some point, simplifying the return_url stuff at least somewhat. However, that is scheduled for some undetermined future time, and the short term fix that I have committed simply consists of removing the button bar from the edit and create confirmation pages, as well as error pages. That should do it, while hopefully not impacting usability significantly. Yngve Svendsen Gnatsweb maintainer Chuck Alexander wrote: > Some time ago I filed gnatsweb PR 507 describing at least one > mechanism by which this can happen. Maybe this is what you're seeing? > > 1. File a PR the normal way and submit it. A confirmation page is > shown. > > 2. On this confirmation page, click the "Create" link at the top of > the page. On the new Create page, look at the page source... note > the return_url defines the PR filed in step 1. > > 3. Fill out the second Create form and submit it. A confirmation > page is shown, which after the refresh period submits the > "return_url", effectively refiling the first PR and displaying a new > confirmation page. > > - Chuck > > Jo A Wahle wrote: > >> >> My users have reported at least three cases where a PR was submitted >> twice. I've captured details about two cases - in one case, the >> duplicate's Arrival-Date is more than 5 minutes after the original, >> and in one case the duplicate's Arrival-Date is one second later. >> We're using gnats 4.1 and gnatsweb 4.00 plus some local enhancements. >> The users are certain they did not submit the same information twice, >> and they didn't use the Back button and submit the same data again. >> Each case involved different users and different categories. I don't >> know whether to blame gnatsweb or gnats, and I'm not sure how to >> start to investigate this, since I can't re-create the problem. Any >> ideas? >> >> -- >> Jo Wahle, Software Engineer, Engineering Support, SAIC >> (Science Applications International Corporation ? www.saic.com) >> Contractor to U.S. Geological Survey ? EROS, Sioux Falls SD 57198 >> (Earth Resources Observation and Science ? eros.usgs.gov) >> >> ------------------------------------------------------------------------ >> >> _______________________________________________ >> Help-gnats mailing list >> [email protected] >> http://lists.gnu.org/mailman/listinfo/help-gnats >> >> > _______________________________________________ Help-gnats mailing list [email protected] http://lists.gnu.org/mailman/listinfo/help-gnats