lubuntu-qa team mailing list archive

[Jonathan Marsden <jmarsden@xxxxxxxxxxx>]

On 06/18/2013 05:59 PM, Ali Linx (amjjawad) wrote:

>>>> I'd much prefer a test using the default install type.  ...

>>> With all due respect, this will make no difference whatsoever,
>>> IMHO. What this has to do with RAM Usage and zRAM and stuff like
>>> that?

>> It is possible that one path through the installer sets up the
>> manually created swap partition sooner, or differently, than the
>> path that automatically partitions the whole disk.

> IMHO, we need an expert who can tell us that :D

Or maybe a tester willing to read the code before stating "this will
make no difference whatsoever" -- as you did earlier! You have a B.Sc.
in Computer Science (just like me -- well, officially mine is B.Sc. in
Computing and Information Systems), so you are clearly capable of
reading that code if you really wanted to.  Just as I am.

> IMHO, testing will never prove or show us that.

Sure it could.  If one code path delays use of swap, and so ends up
using (say) 32MB more RAM than the other, then you could find that a
256MB machine would install fine on one code path, and run out of RAM on
the other.  No expert needed.

> I understand what you are saying, but for someone sitting and
> watching what is happening on a monitor, won't notice or can tell
> what is going on deep inside.

You won't notice the difference between a successful install and a
failed one?  I *hope* you would :)

>> Second, if you are sure "it makes no difference whatsoever", why
>> would you choose the manual partitioning way -- it needs more time
>> and more mouse clicks, for what you say is "no difference
>> whatsoever"?!

> Because, I have two HDD (please check the report)

I know... I didn't make a fuss about that, but testing with only one
connected to the system would be a more common test configuration.

> ... and I have Data on the first HDD (BUT NO SWAP Partition) and I
> just did the manual approach and to be honest, as I said, 4 tests
> were done via "Automatic" approach so, thought some kind of a change.
> But, tell you the truth? I knew it you would disagree :D I was right
> :D but yet again, because I have two HDD, I just wanted to do the
> manual way.

I want a documented and repeatable set of tests, that test what we are
trying to test.  No more, no less.

What are we testing for?  To determine RAM usage requirements of the
default Lubuntu install from an Lubuntu 13.10 desktop i386 image, used
by a novice user coming from Windows XP on a normal older PC machine.

Why documented and repeatable?

So that anyone reading your report can do the exact same test on their
hardware, and get the same results.  That's what testing is all about if
you want it to be useful.  That's why scientists need to document methdo
as well as results, and why duplicating the results of an experiment
someone else did is often considered very useful science!

In the computing world, bug reports need "steps to reproduce", and tests
need to be documented in sufficient depth that others can repeat the test.

If you want to do whatever tests make you happy, with whatever
variations you feel like at the time, well, that's up to you... but it's
less helpful to the one single goal we are trying to reach regarding
testing RAM usage requirements of the default Lubuntu install from a
desktop i386 image, for someone coming from Windows XP on a normal older
PC machine.

>> Don't lose sleep doing Lubuntu testing :)

> How can I sleep and you are asking me for more tests? :P :P :P hehe

No, I'm *not* asking you for more tests.  I'm just saying that the one
you did is probably a little less useful and less repeatable than the
test that I *thought* you were going to do.  As far as I know, I did
*not* ask you to do the test again.  I really was suggesting that you
don't replace sleep with Lubuntu testing!

> Anyway, no promises here but if I get a chance, I will do it, just
> for you despite the fact I'm not convinced at all.

Thanks (remember, you are volunteering this, I am not asking!).  It
probably will make no difference to the result.  But IMO it will be a
much stronger, more useful result.

Jonathan