Apache OpenOffice (AOO) Bugzilla – Issue 7901
auto-retry (without error message) when saving a file to a file system fails
Last modified: 2013-02-07 22:34:38 UTC
When the save button is pressed and the filesystem is not reachable at that moment, no error message should be answered. Insted the Application should wait for 5 seconds and then retry. If access still failes, an error message may be issued. Instead of 5 seconds, a configurable fileover_time variable could be used by all apps. This is essential, because without a retry after a certain time it is very hard to deliver high availability. So many apps are innecsesarily and sadly made unusable for HA environments this way, all email cliens, all biff clones, all office programs, all graphic apps....Its realy sad. Many even crash in this situation (OO for Windows for example), but thats bug 7900. Regards Patrick ---
*** Issue 7904 has been marked as a duplicate of this issue. ***
Patrick, I set this to ENHANCEMENT instead of defect - what you request is a feature in my opinion. Additionally, I change the title - "General Design Error" does the reader not tell anything, and a title is supposed to get a quick impression what this bug is about. Yet more additionally, I change the component/QA contact, as this is a framework issue, not an UI one. Even yet more additionally, I wonder what kind of system allows the file system to be temporarily not present - this sounds strange to me ...
From issue 7902 > I set this to ENHANCEMENT instead of > defect - what you request is a > feature in my opinion. A message, that makes the user believe, something is not working properly, even though this is not true, is a defect. A Failover Situation is not an Error from the user perspective, it is an event only relevant to the Administrator. The maximum, the user should be informed about is, that a slight delay in services is about to happen. > Additionally, I change the title - "General Design Error"# > does the reader not tell anything, and a title is supposed > to get a quick impression what this bug is about. Sorry, my fault. > Yet more additionally, I change the component/QA contact, > as this is a framework issue, not an UI one. ok. But adding just a Retry Button to the Error Message with a recommendation to wait 15 seconds would be a great improvement and and easy to implement. > Even yet more additionally, I wonder what kind of > system allows the file system to be temporarily not > present None that I konw of. The question is, who will be first, the OSS World or the MS World. > - this sounds strange to me ... I would be interested in a "non strange" HA concept for this sort of problem. A HA concept for recovering from failures (even unnoticable to the user) is easy in many cases and works even without changing samba specifications for example. A HA concept, that simply forbids failures to happen is difficult, expensive, hardwaredependent and incompatible with almost everything. Regards, Patrick
*** Issue 7902 has been marked as a duplicate of this issue. ***
> A message, that makes the user believe, something is not working > properly, even though this is not true, is a defect. A Failover > Situation is not an Error from the user perspective, it is an event > only relevant to the Administrator. The maximum, the user should be > informed about is, that a slight delay in services is about to happen. well, I would think it is. At the very moment the file should be saved, no file system was available. Yes, it's a technical point of view :). But I always suspect magic, and what you initially suggested is Magic. I think this because the reason for the error is not clear - waiting 5 seconds would mean waiting 5 seconds in other error cases, too, and this is something which I definately do not want from my office suite. In _my_ daily work, I much more often encounter errors (such as missing permissions or whatever) other than a _temporary_ unavailability. And I do not want to wait 5 seconds until my program tells me. (and no, I do not believe that the magic can be improved to a point where it reliably distinguishs this - I am a skeptic :). The retry button may be reasonable, and indeed it may not be too difficult to implement. For the other things you said: Sorry, I do not know what HA means :). I admit my techncal knowledge there is limited, I just looked at this from my "do-everything-on-my-big-fat-desktop-running-on-a-DOS-successor-machine" perspective :).
Its no good to use to much acronyms. Sorry, with HA I meant High Availibility (its hard to type, so for the rest of this message, Ill stick to HA). The Situation I was concerned about regarding OpenOffice was, that a server goes down and a different server with mirrord disk content takes over. This is difficult to achieve in less than 5 seconds at the actual stage of technoloy. Does somebody know, if IPV6 will improve this? HA is a point, where Linux can beat Windows, when done with Software. At this moment, most commercial solutions are in Hardware and very expensive. Only the low prices for Network Devices make Software based HA on cheap Systems with cheap IDE disks feasable. 2003 shoud be THE big year for Linux HA. Not much has to be done for it in client Software, but everybody must do a little, because when Softwarebased, there is no such thing like a HA Server. The whole System is HA or not. It might be a good Idea to have a systemevent "Failover in process", but at the moment this is not an OpenOffice issue. While it is not there, you may be right, that it makes no sense to try to find out the reason for a file system access problem. >The retry button may be reasonable, and indeed it may not be too >difficult to implement. After your arguments, I now think, that this is the best solution for the nearer future. --
Reassigned to Bettina.
Despite the lack of activity on this, confirming as a new enhancement issue.
To grep the issues easier via "requirements" I put the issues currently lying on my owner to the owner "requirements".