Re: [libreoffice-users] Re: increase testing to improve quality

On 01/03/2012, NoOp <glgxg@sbcglobal.net> wrote:
> On 02/29/2012 03:14 AM, e-letter wrote:
> > On 29/02/2012, NoOp <sniprudeemailquotefrome-letter-fixyourclient> wrote:
> > > > There is a web page which describes manual testing:
> > > > http://wiki.documentfoundation.org/QA/Testing/Manual_Tests. Clearly,
> > > > there are insufficient testers...
> > >
> > > Clearly:
> > > <quote>
> > > IMPORTANT: Obsoleted by Litmus
> > >
> > > Litmus is a useful tool for organizing manual tests. Please, help us to
> > > migrate all these test cases there and remove them from this wiki page.
> > > </quote>
> > >
> > > So why even point 'users' to that page?

As stated previously, historical; only recently read about this
following comments by users about initial use of version 350.

> > The initial purpose is to promote the existence of QA testing, both in
> > terms of the history (the wiki) and the future (litmus).
>
> And if the manual test is obsolete & LO wish to migrate everyone to the
> litmus test page instead you think it's a good idea to direct users on
> this list to that page? Seems a bit odd to me.

It's about reading the information, the wiki page was useful to learn
what happened previously.

> > > Cool. Have you *actually* installed 3.5.1rc1 (as noted in the litmus
> > > test)?
> > > Description:        Test run for *3.5.1 RC* regression testing. Please use
> > > *3.5.1 RC* build to test the cases in this run.
> >
> > No, there are no new features of personal interest.
>
> Well you've got me there... You are asking everyone on this list to hop
> on over & perform pre-release testing & yet you've not bothered to
> install & do the same? Amazing.

No, asking those complaining about performance of the latest version.
I am almost never interested in new features (apart from fodf
recently).

> > > If you did (and you'd need to find it someplace other than:
> > > http://www.libreoffice.org/download/pre-releases/ and (currently) only
> > > on http://dev-builds.libreoffice.org/daily/
> > > you'd find that the install (at least on a debian linux system) smokes
> > > your existing 3.5.0 install (remember 3.5.0 is the released version).
> >
> > I don't fully understand, but perhaps this is a serious bug?
>
> No. It's not a bug at all. It is a fact of testing. If you actually
> actively did what you are flogging in this thread (pre-release testing)
> you'd be well aware of the issues with installing test versions on
> existing systems. To promote otherwise on *this* list is both
> disingenuous and dangerous.

Read the title, it's about improving software qa. As stated elsewhere,
pre-release testing must never be implemented on important machines.
There is no danger in telling intelligent people to use software that
has passed suitable qa testing and most importantly, be aware of the
need for qa.

> > > So, while I (we?) appreciate your efforts to get users on this "user"
> > > list as pre-release testers, I'd be overly cautious about asking "users"
> > > on this list to perform litmus tests... At least not without explaining
> > > in *detail* what is required, and how to avoid existing LO standard
> > > installations being blown away by a pre-release version.

Please indicate a web page publishing these details and users will know.

Goes back to previous comments: who would seriously test software on a
computer with important data??? Noone is asking/telling users
anything, except to be aware of testing and consider contributing.
Firstly, the assumption is that users are not aware of qa testing,
which needs to be changed before LO improves.

> You've already stated that you've not installed/tested/tried 3.5.x
> yourself ("No, there are no new features of personal interest.") and are
> unaware that the "RC" does not install as a "dev" version and installs
> over the top of a pre-installed 3.5.0 "standard" version. So... are you
> sufficiently intelligent enough to understand that your "advise" made no
> mention of this?

Exactly why the information you wrote should be promoted as was suggested.

> > Any competent tester should have a separate machine.
>
> Right... Do you? I seriously doubt that.

Which is why such high risk tests are not performed!

> "Any competent tester" would have tested on his/her own system *first*
> (with at several OS's) before posting on this list.
> Did you?

Again pre-release software is not of personal interest, so never use
it. Others have the right to be interested...

> Note: If you (or anyone else) is still reading; I highly advocate
> testing pre-release versions of LO and AOO whenever possible. However I
> do not advocate posting "invites" on an LO (or AOO/OOo) user list
> recommending that the "users" on the list jump on in and "test" an
> alpha/beta/RC version of LO (or anything) without *first* advising the
> users/testers of the issues (including data loss/profile loss etc) in
> doing so blindly.

I respectfully disagree and consider users sufficiently intelligent to
evaluate the risks themselves.

> @e-letter: I'm not attempting to dissuade you (and others on this list)
> from from promoting/asking others to assist in testing alpha/beta/RC
> versions of LO. I *am* attempting to dissuade you from doing this
> without first: 1) doing it yourself, 2) knowing & explaining the
> benefits & hazards of doing the tests, 3) where & how to report what
> you've done/tested.

1: Again, I am aware of the risks and I expect users to be similarly
capable to evaluate risk to their own satisfaction.
2: Those responsible for management of qa testing should have
published such information.
3: If there are no instructions...

Assuming you have performed this tests, you should question where
these factors you described are published clearly for users.

> I would add: 4) ensure that any particular "user" on this list be
> prepared to *not* attempt to engage in any dev LO devs/lists regarding
> issues they find with testing these "pre-release" versions, particularly
> those that may involve developers as they will be greeted with both
> animosity and insults in the process. You'll then be asked for a
> software patch & solution to the problem that you've inocently brought
> up on the dev list, and then be sent to dev-hell because: 1) you've
> dared to bring up a user issue on the dev list, and 2) did not provide
> the 'devs' with a software patch to fix *your* problem.

Interesting, a common expression of programmers' disdain for those
"users" actually using the software. A current user may become a good
tester in the future. When you read the post on the qa web site, the
question is posed, how to formally consider usability?

> You'll find the same if you attempt to venture into the LO UI list
> arena... See:
> https://bugs.freedesktop.org/show_bug.cgi?id=46073
> and

QA testing should have prevented the software being released, had such
a suitable manual test be included in a formal qa procedure. This is
one weakness of the "regular" software version release schedule,
ignoring whether bugs have been removed.

> the changes aren't available in 3.5.x (where the bug was filed to begin
> with), we'll take the Ubuntu linux approach and mark the bug as 'fixed'

Had to laugh at the ubuntu criticism: is that really what they do???
No wonder those "anti-ubuntu" comments are increasing throughout the
web.

> http://wiki.documentfoundation.org/Development/Use_of_MailList
>
> <quote>
> Do not mail to the list for:
>     Reporting bugs, or feature requests unless - you are actively
> soliciting feedback on how best to fix them yourself
>         Reporting bugs is explained here:
> http://wiki.documentfoundation.org/BugReport
>         User questions and general feature discussions can be done on
> the lists users@global.libreoffice.org and discuss@documentfoundation.org
>     This mailing list is particularly not a good place to report your
> views on the general quality, completeness etc. of the software - unless
> connected with actively working to improve the code.
> </quote>

QA is as important as bug reporting; the observation that users are
encouraged to submit bugs, but _not_ to contribute to improving
quality, is a weakness. More users becoming aware of testing,
hopefully becoming sufficiently interested in testing, would improve
software quality.

> If this 'user' list would like examples of the above I'll be happy to
> post them.
>
> Anyway, good luck with sucking in unsuspecting *users* on this list to
> download (over 130MB of data) of some alpha/beta/RC bits, fire it up and
> have it overwrite their existing LO/OOo (you do realise that the "RC"
> will do this correct rather than install into a 'dev' opt & profile?)...
> nah nevermind - you apparently didn't have the intelligence nor the test
> machine to test before popping up with your 'advise'.

That is your incorrect opinion...

-- 
For unsubscribe instructions e-mail to: users+help@global.libreoffice.org
Problems? http://www.libreoffice.org/get-help/mailing-lists/how-to-unsubscribe/
Posting guidelines + more: http://wiki.documentfoundation.org/Netiquette
List archive: http://listarchives.libreoffice.org/global/users/
All messages sent to this list will be publicly archived and cannot be deleted