I'm not giving up yet, even though I've spelled out this preference for Maverick, Lucid, Karmic, Jaunty and probably earlier releases as well.
Not everybody has the money to buy a new machine every few years. Some people just don't see the need to replace perfectly functioning hardware just to be able to run the latest and greatest. "OK, but then just don't run the latest and greatest" you say? That is of course an option but it is not ideal in many ways. It should also not be necessary which is why I keep on coming back to this issue.
In the earlier stages of the Maverick test cycle performance on my Thinkpad T23 machines was quite bad. Memory consumption went through the roof - not something you want on a machine with 768 MB, less even on the one with 'only' 384 MB. Just booting the thing to the desktop used up more than half of the available memory. This has gotten better lately but it can and in my opinion should be improved even further.
Why? If your box has 4 GB you might not see any reason to put effort in reducing memory footprint. But... wait a few years and that 4GB will seem as cramped as my 768 MB does at the moment. When simple panel applets start using 256 MB because they are written in some all-singing and dancing object relational prototype actor framework reflective introspective genuflective language on rails or with wings for all I care you will start to wonder where that memory went. Or maybe you won't because by then you will have bought a new box with 64 GB so who cares...
Well, I care.[INDENT][SIZE=3][COLOR=Silver]<p class="getoffmylawn">[/COLOR][/SIZE][SIZE=3]I learned to ride a bicycle on something which closely resembles the one I ride now. The Model A Ford in my father's garage is not that dissimilar from what I see on the road every day. My Ural motorbike may look old-fashioned but it is not that different from more modern bikes. The computer I'm using right now - one of those T23s - is much faster and has much more memory than the Commodore 64 I bought in the early 80's... but in many ways it is not *that* different.[/SIZE][SIZE=3][COLOR=Silver]</p>[/COLOR][/SIZE] [/INDENT]That 64 KB which seemed like a lot when the C64 was new is now more than gobbled up by the stack of a simple applet to adjust sound volume:
bfe8b000-bfeac000 rw-p 00000000 00:00 0 [stack]
Size: 136 kB ...Of course that old machine was mostly programmed in assembly which was less than ideal. Fortunately we have alternatives to this now, from C (structured assembly :) and its offspring all the way to Haskell and beyond. Using these in lieu of assembly does use more resources but that does not matter as we have more.
That does not mean those resources should be wasted though... How?
By not using memory hogs in the base install. By avoiding the use of continuously running interpreted code - if it needs to be running all the time use a compiled language when feasible. By going over the current code base and looking for ways in which it can be made more efficient, possibly by combining disparate applets in one code base. It should not take several megabytes to display a small icon in a toolbar! My whole window manager (Xmonad) takes half the amount of memory compared to the Network-Manager applet!
Just stating a preference does not make things happen. I therefore propose a project to see if we can tame the hogs:
[SIZE=5]Project Hog Skinner[/SIZE]
Aim: to reduce resource consumption of the base install of one or more Ubuntu flavours. Who can participate: anyone who has the required knowledge and experience to identify and resolve resource hogs in the base install. How: contributors can 'adopt' a program, range of programs or package. This adoption is not exclusive, more than one contributor can work on a single program. Three months before release we try to combine the efforts of all contributors. Timeline: starting NOW