23/10/2006 4:16:17 PM|rosetta@home|Reason: Requested by user
23/10/2006 4:16:17 PM|rosetta@home|Requesting 604800 seconds of new work, and reporting 2 completed tasks
23/10/2006 4:16:23 PM||A new version of BOINC (5.6.5) is available for your computer
23/10/2006 4:16:23 PM||Visit http://boinc.berkeley.edu/download.php to get it.

This has happened on more than one machine, the new version is not available anywhere?

If memory serves correctly, the message has appeared after aborting a WU and then reporting it back...

Edit: All machines this has occurred on are running 5.4.11, Win XP SP2 (either pro or home).

---

Boincs' "all versions" php

NOTE: 5.6.5 is still alpha, use at own risk etc. You will occasionally see "bad libcurl argument" message in "messages tab", this has already been fixed in the next version and doesn't really harm anything. A few other minor issues have also been addressed, but overall, it seems to work fine for me.

Boincs' "all versions" php

NOTE: 5.6.5 is still alpha, use at own risk etc. You will occasionally see "bad libcurl argument" message in "messages tab", this has already been fixed in the next version and doesn't really harm anything. A few other minor issues have also been addressed, but overall, it seems to work fine for me.

Thanks, but why is the server advertising an alpha?

---

Thanks, but why is the server advertising an alpha?

The release of the new gui version is some time off and I think Dr. Anderson feels that this part of the new features is too important to the participants to wait. If the current "recommended" release works for you and you don't want "cpu throttling" then I'd recommend you stick with the "recommended" version. The following was posted by Dr. Anderson to the "Boinc Projects" mail list 3 days ago.

A few weeks ago we added a general preference for CPU throttling
(useful for controlling CPU heat).
This preference is enforced only by the 5.6 client and later.
There have been lots of complaints because the preference is shown
(e.g. on the SETI@home site) but is not enforced by the current client release.
So:

1) I added text to the preferences page saying that CPU throttling
is enforced only by 5.6+ client
(check out html/inc/prefs.inc to get this change)

2) The standard download page on the BOINC site (download.php)
now has a link ("All versions") to a page that that now
really shows all versions, including development versions
(no need for the secret "dev=1" parameter).
This lets people use CPU throttling if they really want it.

-- David

That still doesn't answer why on intermittent times, a scheduler response from Rosetta ONLY has the whole "There is a new version available" message....

---

Think of it like a billboard which says "your next new car is available". It doesn't say "you MUST upgrade" or anything like that. One of the biggest issues for returning participants was the heat boinc/seti caused. If you read the Q&A boards at Seti you'll see many people would rather limit cpu useage than clean their puters.

There have also been participants who expressed a desire for boinc to inform them when there was an upgrade available, rather than to have to visit the boards to find that out.

Think of it like a billboard which says "your next new car is available". It doesn't say "you MUST upgrade" or anything like that. One of the biggest issues for returning participants was the heat boinc/seti caused. If you read the Q&A boards at Seti you'll see many people would rather limit cpu useage than clean their puters.

There have also been participants who expressed a desire for boinc to inform them when there was an upgrade available, rather than to have to visit the boards to find that out.

They're even looking into autoupdating now as well (at long bloody last).

---

Team mauisun.org

Ahhh, fair enough, thanks :)

---

...There have also been participants who expressed a desire for boinc to inform them when there was an upgrade available, rather than to have to visit the boards to find that out.

They're even looking into autoupdating now as well (at long bloody last).

The latest version of boinc -- though tested -- isn't always bug-free. An auto-update feature can become a quick way to spread the newest bugs around. I'm using 5.4.9 instead of 5.4.11 and content with it. But I have one machine, someone with a farm would appreciate an auto-update I'm sure.

...There have also been participants who expressed a desire for boinc to inform them when there was an upgrade available, rather than to have to visit the boards to find that out.

They're even looking into autoupdating now as well (at long bloody last).

The latest version of boinc -- though tested -- isn't always bug-free. An auto-update feature can become a quick way to spread the newest bugs around. I'm using 5.4.9 instead of 5.4.11 and content with it. But I have one machine, someone with a farm would appreciate an auto-update I'm sure.

So would WCG/IBM ;-). The main reason for them not doing it (auto-update) was security reasons (or lack of knowledge and confidence with possible auto-update security risks).

The reason I asked Rom that question is that newer version often add new features, if projects wish to make use of the new features, like Graphics cards, SSE3/4 etc. Then they need to know that the majority of participants (the fit and forget joe public) will be using the latest versions and so can actually get that information. Until people move to 5.6+ versions people cannot build up states for what Grpahics cards people have decide if it is worth tagetting that platform OR they need to use fpops or a new inteligent memory managment planned for 5.8 series.

Or they discover a large security bug in the program and it need securing quickly. If you cannot get the version to all the people then you got a large problem and may cause a big step back in people maybe trying it out if a large attack happened.

Since the majority of people are fit and forget it very useful.
For people like you and me we may just want a, hey there is a new version
Or you may not want to know at all.
Auto-update always should be an option to enable/disable by the user.

---

Team mauisun.org

...Auto-update always should be an option to enable/disable by the user.

You give good reasons for supporting auto-update. I would be comfortable with the option of enabling or disabling.

I agree with the point made earlier, that it is still odd that an alpha is being so advertised. If an alpha is advertised the message should say so.

...Auto-update always should be an option to enable/disable by the user.

You give good reasons for supporting auto-update. I would be comfortable with the option of enabling or disabling.

and in the same thrain of thought, maybe auto update should have a three/four way choice, not just an on/off

- notify all but no auto update (default)
- install latest alpha
- install latest beta
- notify dev releases but only auto install latest stable release

(for a three way choice, combine alpha/beta into latest developmental release)

My guess is that it would be no more challenging for the users, nor for the programmers, if a multi-way choice were added at this stage.

I personally would strongly welcome auto-update, but only if it was for stable releases (or if that were one of the options). With 11 machines, updating BOINC versions is a pain.

Finally, there are other possibilities: provide ways that the updating of the client can be done by the client on user command from the manager, from boinc_command --update_client, or from an a/c manager.

Just some thoughts - feel free to pass them back to the BOINC programmers if you think any of them merit it.

River~~

---