In my opinion this should be the first project to set the minimum core client version to whatever the current public release is.

I think this because there are alot of members here that are concerned about upgrading the client in the middle of a model. Setting the minimum would prevent older clients from downloading new work without hindering reporting of completed work, and thus the upgrade could be performed with no model on the computer. With the long time frame of the workunits here I do not think it would be too much of an annoyance. Some of the other projects have had to back off on requiring more up to date clients because the server was brought to it's knees by requests from clients below the minimum.

The trigger that got me to post this was I saw another post that seemed to indicate that someone had just gotten a new 4.10 workunit on a 4.05 client.

---

BOINC WIKI

BOINCing since 2002/12/8

Good idea, John. I second that motion.

---

"We have met the enemy and he is us." -- Pogo
Greetings from coastal Washington state, the scenic US Pacific Northwest.

Would this annoy people who do not want to risk it on a nearly complete WU, are away when the model completes, and are only running 1 project?

With 3 hoops to jump through most people will not be annoyed, still...

Having a preference to allow a delay seems too onerous on development teams. Only other option I thought about was an email on completing a model to give congratulations but also to include please upgrade the core client to version 4.19. With this people might decide 'if it aint broke' and ignor the request. So the suggestion may well be the best approach.

---

Visit BOINC WIKI for help

And join BOINC Synergy for all the news in one place.

I too am surprised at how reluctant people are to upgrade, but that is true of most software. There have been mixed messages over whether people <i>should</i> upgrade to the latest BOINC version, and some are also confused over public release versus alpha and beta versions of the software.

If the project were to go down the line suggested by John Keck there are a couple of stipulations I'd like to make.

1. We do have to be sure that the minimum core version really is stable.

2. As Crandles points out, waiting until the end of the run does not suit everyone. We need clear upgrade instructions, and the option for the user of falling back to the earlier version in order to complete the run, if the upgrade does fail for some reason. That is not straightforward, as people could be running any one of a good many versions of BOINC and may need to download an earlier version in order to restore a backup.

We could lose a lot of users if we are not careful. Most people probably don't check their messages or BOINC status, many won't easily understand what is required of them, and some do not find it easy to find help on a topic. CPDN lost many users through a security change to Windows a year ago even though a fix was available in days and all were emailed.

---

...
&gt; required of them, and some do not find it easy to find help on a topic. CPDN
&gt; lost many users through a security change to Windows a year ago even though a
&gt; fix was available in days and all were emailed.
&gt;
No doubt CPDN is still on the spam filter list of some ISPs as well. At least I remember Carl complaining about that.