RE: [flexcoders] Re: Lost commit on data source with polling AMF

[Jeff Vroom]

I am checking into the hotfix process right now.   I have the fix working locally on a simple test case and will run our test suite on this fix tonight.     We do currently have the limitation that you have to “merge” before you “commit” or you get an error.  Theoretically, we could relax this constraint and let you merge after the commit but I think it would hit the same “stale conflict” problem if you ended up merging a change to an item after you had already committed and resolved a conflict on that item.      So you are saying that if you call “merge” followed by an immediate “commit” that you are still running into problems with unmerged messages?   Messages pushed from the server will not be delivered until your code has finished running so if you call merge() immediately followed by commit() the commit should only generate an error if the merge caused a conflict.  In that case, you’d have to resolve the conflict before you can commit anyway (we also require you to resolve all conflicts before you can commit).  If this doesn’t match what you are seeing, let me know.   Jeff   From: flexcoders@yahoogroups.com [mailto:flexcoders@yahoogroups.com] On Behalf Of [EMAIL PROTECTED] Sent: Wednesday, September 13, 2006 8:43 AM To: flexcoders@yahoogroups.com Subject: Re: [flexcoders] Re: Lost commit on data source with polling AMF   Thanks for the responses, how long does it usually take to implement a hotfix? We have been playing around with a datasource that has autocommit and automerge set to false, but it appeared to be almost impossible to issue merge() and commit() sequences without running into problems. Are there any examples/tutorials on how to structure the code for a merge-resolve-commit cycle with auto-commit and auto-merge turned off? We found that sometimes the commit would throw an error because there were pending merges, this despite the fact that a merge was just completed. I presume this arose from a further push that arrived between the resolution of the last conflict and the commit request. Is there a way to make this operation atomic? By that I mean something like a mergeAndCommit() which resolves all merges, locks the data source, and then commits. Kornelius [EMAIL PROTECTED]ups.com wrote on 13/09/2006 02:13:27: > What is happening here is that the latency introduced by the poll is > causing the client to receive the response from its "commit" before a > previously queued pushed message to that client. By the time it > receives the pushed message, it is "stale" - i.e. the previous > version is not accurate anymore. We should be sending those queued > messages to the client before the client processes the response to > its operation. Thus the client would receive notification of the > conflict two times in this case but it would be the same conflict so > you would only have to resolve it once. > > The only workaround I can think of right now is to resolve that > second (stale) conflict using "acceptClient" without reporting it to > the user. The problem though is knowing when this is a stale > conflict versus a real new conflict. There is a timestamp in the > message which theoretically could be used to detect that but it is > pretty tricky so I'm not sure I'd recommend trying this workaround. > I am working on a fix now... once it is ready we could investigate > the possibility of turning it into a hotfix. > > Sorry for the problems and thanks for reporting this issue. > > Jeff > > > > > > -- > Flexcoders Mailing List > FAQ: http://groups.yahoo.com/group/flexcoders/files/flexcodersFAQ.txt > Search Archives: http://www.mail-archive.com/flexcoders%40yahoogroups.com > Yahoo! Groups Links > > > > > > > > > > > > __________________________________________________________ > This e-mail has been scanned for all viruses by Star. The > service is powered by MessageLabs. For more information on a proactive > anti-virus service working around the clock, around the globe, visit: > http://www.star.net.uk > __________________________________________________________ __._,_.___ -- Flexcoders Mailing List FAQ: http://groups.yahoo.com/group/flexcoders/files/flexcodersFAQ.txt Search Archives: http://www.mail-archive.com/flexcoders%40yahoogroups.com SPONSORED LINKS Software development tool Software development Software development services Home design software Software development company Your email settings: Individual Email|Traditional Change settings via the Web (Yahoo! ID required) Change settings via email: Switch delivery to Daily Digest | Switch to Fully Featured Visit Your Group | Yahoo! Groups Terms of Use | Unsubscribe __,_._,___