I'm still trying to figure out what is going on with my dual core only using one core when it always used both before (my general prefs on BAM! have it at 1 CPU, but this has somehow never been a problem until now). Here is what happens when I try to sync with BAM! As you can see below, it runs CPU benchamrks, tells me that my number of CPUs has changed from 1 to 2, then immediately runs another set of CPU benchmarks and concludes that the number of CPUs has changed from 2 to 1. Hmmmm. Any suggestions? I have put this same message on the rosetta and BAM! boards to no avail.

Here are the messages:


3/12/2007 6:25:27 PM||Contacting account manager at http://bam.boincstats.com/
3/12/2007 6:25:27 PM||Account manager: BAM Host-ID: 29811
3/12/2007 6:25:27 PM||Account manager contact succeeded
3/12/2007 6:25:27 PM||General prefs: from http://bam.boincstats.com/ (last modified 2007-03-12 18:25:35)
3/12/2007 6:25:27 PM||Host location: none
3/12/2007 6:25:27 PM||General prefs: using your defaults
3/12/2007 6:25:27 PM||Number of usable CPUs has changed from 1 to 2. Running benchmarks.
3/12/2007 6:25:27 PM||Running CPU benchmarks
3/12/2007 6:25:27 PM||Suspending computation - running CPU benchmarks
3/12/2007 6:25:45 PM||General prefs: from http://bam.boincstats.com/ (last modified 2007-03-12 18:24:06)
3/12/2007 6:25:45 PM||Host location: none
3/12/2007 6:25:45 PM||General prefs: using your defaults
3/12/2007 6:25:45 PM||Number of usable CPUs has changed from 2 to 1. Running benchmarks.
3/12/2007 6:26:29 PM||Benchmark results:
3/12/2007 6:26:29 PM|| Number of CPUs: 1
3/12/2007 6:26:29 PM|| 1921 floating point MIPS (Whetstone) per CPU
3/12/2007 6:26:29 PM|| 4085 integer MIPS (Dhrystone) per CPU
3/12/2007 6:26:30 PM||Resuming computation
3/12/2007 6:26:30 PM||Running CPU benchmarks
3/12/2007 6:26:30 PM||Suspending computation - running CPU benchmarks
3/12/2007 6:27:31 PM||Benchmark results:
3/12/2007 6:27:31 PM|| Number of CPUs: 1
3/12/2007 6:27:31 PM|| 1946 floating point MIPS (Whetstone) per CPU
3/12/2007 6:27:31 PM|| 4123 integer MIPS (Dhrystone) per CPU
3/12/2007 6:27:31 PM||Resuming computation

---

Wait until Willy answers on the BAM boards. I already pointed him to your thread on Rosetta earlier. I feel it's a BAM problem first... not a BOINC problem.

---

The reason it's not working and the reason you haven't figured out why it's not working is because you've fallen prey to the "logic" that says you must use every silly bell and whistle every silly programmer trying to be somebody can dream up. Forget about BAM, it will only make life more complicated for you than it needs to be.

Remember KISS... keep it simple stupid. Put your preferences in a file named global_prefs_override.xml in your BOINC folder and be done with the problem once and for all. If you want to continue having problems then just keep doing it the way you're doing it now.

Saving preferences in BOINC's Simple View will also screw up your preferences and cause you nothing but grief and sorrow.

Remember KISS... keep it simple stupid. Put your preferences in a file named global_prefs_override.xml in your BOINC folder and be done with the problem once and for all. If you want to continue having problems then just keep doing it the way you're doing it now.

Saving preferences in BOINC's Simple View will also screw up your preferences and cause you nothing but grief and sorrow.

That's a contradiction, BOINC's Simple View uses global_prefs_override.xml.

Remember KISS... keep it simple stupid. Put your preferences in a file named global_prefs_override.xml in your BOINC folder and be done with the problem once and for all. If you want to continue having problems then just keep doing it the way you're doing it now.

Saving preferences in BOINC's Simple View will also screw up your preferences and cause you nothing but grief and sorrow.

That's a contradiction, BOINC's Simple View uses global_prefs_override.xml.

Indeed it does use global_prefs_override.xml, just doesn't use it nicely, it appears. Apparently, Simple View allows only a subset of the preferences tags BOINC supports. If you create a global_prefs_override.xml manually with a text editor and include tags Simple View does not support and then use Simple View to modify a preference it will delete any tags it does not recognize when it updates the global_prefs_override.xml. It happened to me a few days ago. Check it out for yourself, I would love to be proven wrong. IIRC the <ram_max_used_busy_pct> tag is one that is not supported and there are other tags too. So the "grief and sorrow" I experienced is that I thought I had turned on Preference X but Simple View deleted it which effectively turned it off for me. Was very confusing for a while until I realised what Simple View was doing. I guess the way around the grief and sorrow is... if you use a global_prefs_override.xml then either use Simple View to edit it or edit it with a text editor but don't mix the 2 methods.

Any preferences not in the global_prefs_override.xml file made by the simple GUI are read from global_prefs.xml

But yes, any preferences you set in a hand written global_preferences_Override.xml file are overwritten with the Simple GUI's override file. The Simple GUI isn't for the professionals. It's for the people starting with BOINC and expecting to find some basic preferences setting ability.

There will be additional preferences added in newer versions. These were just the main ones.

---

But yes, any preferences you set in a hand written global_preferences_Override.xml file are overwritten with the Simple GUI's override file. The Simple GUI isn't for the professionals. It's for the people starting with BOINC and expecting to find some basic preferences setting ability.

Whatever the target audience, that's just plain bad programming - a bug, in my book.

If the simple GUI wants to 'own' a settings file (i.e. have permission to overwrite/delete things put there by some other process), then it should have its own file - it shouldn't use a file designated for professionals to use in a different way.

OTOH, when I started writing this sort of code (about Windows 3.1 !), we used .INI configuration files for this sort of thing. The Windows API had - and still has, up to XP SP2, not tried it on Vista yet - ProfileFile calls which allow you to read and set individual items in the file, without disturbing the rest of the contents. Surely by now someone's got a library which can do the same thing with .XML files? It'll make things much more maintainable in the future if they would put the underlying file handling on a firmer footing first, before adding the extra bells and whistles.

---

OTOH, when I started writing this sort of code (about Windows 3.1 !), we used .INI configuration files for this sort of thing. The Windows API had - and still has, up to XP SP2, not tried it on Vista yet - ProfileFile calls which allow you to read and set individual items in the file, without disturbing the rest of the contents. Surely by now someone's got a library which can do the same thing with .XML files? It'll make things much more maintainable in the future if they would put the underlying file handling on a firmer footing first, before adding the extra bells and whistles.

Of course there are XML libraries, and lots of them, with many different APIs. There is even one API (SAX if I remember correctly) that lets you read big XML files without having to load the whole tree into RAM. And XPath is great to look for specific tags, the basics are the same as a directory tree, like /html/body/h1.

The problem is that BOINC doesn't use any XML library. To write XML files, they use printf (like fprintf(fp,"<filesize>%d</filesize>\n",file.size);), and to read XML files, they have a self-made parser, which is quite broken (really works only with the files BOINC itself creates). It requires each tag to be in a separate line, to name one flaw. Makes it hard for a 3rd party program to edit BOINC XML files using a proper library, as those libraries keep the original whitespace but don't add any extra when adding tags (and you can also configure them to drop all useless whitespace).

I hear what you say, but that doesn't mean I like the underlying implications :-(

---

I hear what you say, but that doesn't mean I like the underlying implications :-(

I don't like them either, at all... But I have talked to the devs about having a decent XML library, and got answers like "I didn't find any API that I like". Okay, maybe a DOM tree takes a long time to build, and SAX is cumbersome, but look for others then! I think the API their custom parser has is quite horrible...