Linked by Eugenia Loli on Fri 7th Mar 2008 21:38 UTC, submitted by SReilly
Novell and Ximian Open-source pioneer and Novell Vice President Miguel de Icaza Thursday for the first time publicly slammed his company's cross-patent licensing agreement with Microsoft as he defended himself against lack of patent protection for third parties that distribute his company's Moonlight project, which ports Microsoft's Silverlight technology to Linux.
Thread beginning with comment 304017
To view parent comment, click here.
To read all comments associated with this story, please click here.
RE[8]: Interesting
by PlatformAgnostic on Sun 9th Mar 2008 21:47 UTC in reply to "RE[7]: Interesting"
Member since:

Rereading your post made me think of another question...

Why would you run your system with swap off? When you say swap off, do you mean that the OS has no place to page to besides the original executable images for memory that's not in use?

Reply Parent Score: 2

RE[9]: Interesting
by sbergman27 on Sun 9th Mar 2008 22:26 in reply to "RE[8]: Interesting"
sbergman27 Member since:

I mean that swap muddies the waters when reporting memory usage numbers. I always wonder what people mean when they say "Application X uses Y MB of memory". (Windows users, for some reason, seem particularly smug in their confidence in the absolute numbers they report.) Shared memory and swap make it very difficult to report absolute numbers in a meaningful way.

As it happens, I *was* running with swap turned off throughout my whole Epiphany session, by accident. I had it turned off for a completely different experimental reason, and then forgot to turn it back on again. I don't believe that swap is bad and leave it turned off as some people apparently do. As Andrew Morton said, you really don't want megabytes and megabytes of bloatyapp floating around in your ram.

With swap turned off, "resident - shared" actually means something. If there is stuff in swap, you really can't know what "resident" truly means. On the other hand, one could make an argument that having swap turned on, and measuring when the system is under memory pressure, might give a more meaningful representation of the true cost of the app under real life conditions. If it has a lot of stuff that the OS can just swap out once and forget about, that lowers its impact.

At any rate, I always feel conspicuous when I report memory usage numbers, because the real answer lies out there in the gestalt somewhere.

Edit: Oops! I was responding as if you had replied to this post I just made a few minutes ago:

But aside from the Epiphany reference, I think it is still appropriate.

Edited 2008-03-09 22:35 UTC

Reply Parent Score: 2

RE[10]: Interesting
by PlatformAgnostic on Mon 10th Mar 2008 01:14 in reply to "RE[9]: Interesting"
PlatformAgnostic Member since:

That's why I used the phrase 'reference set.' Apps can take tons of memory, but only really touch a little bit of it. "Working Set" is probably the NT equivalent of what you called "Resident Memory." You're right about how to measure the true memory cost of a program... force the OS to drop all of its pages and see what gets pulled back in by the program.

This is really what people should be concerned about when they complain about the bloat of a program. It might have tons of code (to support many features), but if it is well-designed and modular and compiled with a good profile-guided toolchain, the actual cost of all those features is not too bad unless someone is using everything at once.

Reply Parent Score: 2