← Back to team overview

lubuntu-desktop team mailing list archive

Re: Fw: Dropping i386 non-PAE as a supported kernel flavour in Precise Pangolin


On Fri, Jan 6, 2012 at 4:57 AM, amjjawad HOOHAA <amjjawad@xxxxxxxxx> wrote:

> Hi,
> I went through the whole thread (37 emails so far) and noticed very simple
> thing that I already have noticed long time ago (I know what most of you
> will say) which is: we do *NOT *have accurate database of Lubuntu Users
> in general and those who are "still" using OLD machines and I mean VERY old
> here.
> I like such kind of discussion and I tend to agree with some opinions and
> disagree with others but after all, this is very healthy activity
> (discussion) but ... as always ... the real question is: isn't it pointless
> to discuss such thing without REAL figures? some guys already pointed that
> out, I guess.
> Long story short, my simple and humble opinion is: we need a group of
> people who do a survey to obtain such figures. No, I have no smart ideas of
> HOWTO do that but if we could discuss about this, that would be VERY
> helpful. Perhaps we are wasting our time here and there are only 2-3 users
> with such old machines? for me, Lubuntu 10.04 didn't work on this:
> http://ubuntuforums.org/showthread.php?t=1590614 while Mint LXDE 9 did.
> Both may be using the same kernel but I'm not really sure why Lubuntu
> didn't work. Also, perhaps they are many users that will be affected with
> that drop? so we need some figures.
> Just a thought I'd like to share with all, as usual :)
> Thanks!

The problem with a survey is that it will never be representative, since it
is opt-in, and even more unrepresentative based off of where it's
publicized (i.e. a poll of lubuntu-desktop subscribers won't get you all of
the non-pae cpu users). The only fool-proof way of getting this data is to
automatically submit it from users' computers, but that's a privacy issue,
and still limits to internet users and those who do package upgrades
regularly (assuming the submission method results from a package upgrade).

I can offer you this though [1]:
21:35:54 <kees> so, as for data, with bdmurray's help, I downloaded 7271
x86 cpuinfos from bug reports on LP about a year ago. 336 of those were
21:36:12 <kees> so that's just under 5%

I think it's also not how many people are using non-pae cpus, but whether
or not Lubuntu is expected to be able to run on non-pae cpus. For example,
I would expect Lubuntu to run on the 400MHz FSB Pentium M CPUs [2] (only
the 400MHz FSB models don't support PAE [3]); in fact, I'd even argue that
those CPUs are part of the intended audience. And no matter how many users
there are, based off of a QA perspective, if I say that Lubuntu is designed
for a particular kind of machine, it better be able to run on that machine,
plain and simple.

Also, the non-pae kernel is still supported in Precise. All we'd have to do
is change the default kernel on i386.


[2]: http://en.wikipedia.org/wiki/List_of_Intel_Pentium_M_microprocessors
[3]: http://en.wikipedia.org/wiki/Physical_Address_Extension

Follow ups